mirror of
https://github.com/paperclipai/paperclip.git
synced 2026-11-12 06:10:44 +01:00
1b7fe33432016b3b27ccdfbf7b8944da4b40726d
2097
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Dotta
|
1b7fe33432 |
fix(workspaces): preserve import verification across process restarts (#20311)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - Online import preserves old local files before the web server can use a disposable filesystem. > - A parent import must authenticate saved child archives before it can omit their directories. > - The operation journal preserves import identity, but completed child checks existed only in process memory. > - A container restart therefore repeated downloads, decryption and restoration that had already succeeded. > - This pull request saves a versioned verification receipt on each immutable child snapshot. > - A replacement process can reuse those checks while it reacquires locks and verifies current local files. ## Linked Issues or Issue Description Refs #19854 for verification reuse within a live source fence. Refs #19873 for cohort import. Refs #20298 for the separate remote-hold admission fix. Searched related workspace verification and restart PRs; none provides durable child verification. **What happened?** Replacing a server container interrupts a long online import. The original operation and snapshot survive, but a later invocation downloads, decrypts and restores completed child archives again. The current cache ends with the source-lock owner. **Expected behavior** Completed immutable checks survive process and scratch loss. Resume keeps the same operation and snapshot, reuses only correctly bound verification, and publishes once. Current local files and Git state must still be checked under fresh source locks. **Steps to reproduce** 1. Import two nested child workspaces, then start the parent import. 2. Kill the importer after a child verification finishes. Remove its scratch directory. 3. Resume in a new process. Before this fix, the completed child is downloaded and restored again. **Paperclip version or commit** The regression fails on the previous migration verifier. The final branch includes master at `189fefb99183953be37f0d1084275fbb0f35d6e2`. **Deployment mode** Managed Linux server with PostgreSQL and explicit S3 storage configuration. ## What Changed - Add optional opaque storage namespace identity. S3 binds it to the explicit endpoint, region, bucket, normalized prefix and addressing style. Implicit endpoints retain full verification. - Save a versioned child-snapshot receipt only after canonical crypto, manifest and file restoration succeed and a second HEAD confirms the same strong ETag and size. - Bind reuse to the complete snapshot reference, key material, child lineage, source operation and storage namespace. Require a fresh strong HEAD and canonical validation that the current secret provider can unwrap the saved snapshot key. - Keep current directory/Git capture and source-owner checks on every invocation. Read-only verifier callers do not write checkpoints. - Merge checkpoint metadata under a short row lock without changing parent journal hashes, snapshot retention or update timestamps. Bound reads and writes. Lock timeouts, statement timeouts, deadlocks and serialization conflicts are safe cache misses. Other errors remain fatal. - Extend real PostgreSQL/Git/archive tests with process death, scratch removal, resume, changed bindings and single publication. ## Verification - Baseline RED: two import invocations read each child twice. With durable receipts, each child is read once across both invocations. - Real process regression passed on the final rebased head: SIGKILL the owned importer after one checkpoint, delete all scratch, and resume in a new Node process. It retains the same operation and snapshot, reads each child once across both processes, and records one publication. - 53 focused cases passed in 104.86 seconds on the final production bytes before rebase. They cover existing nested/cohort behavior, namespace identity, changed source/reference/key/lineage/ETag, unknown proof versions, weak validators and bounded row contention. - After clean rebase and stronger scratch/snapshot assertions, the real process case passed again in 18.54 seconds. Server TypeScript, module boundaries and diff checks passed on final head `9da68ee7ce4a778201ef85e21a11c49eeea7bc16`. Required CI supplies final-head broad coverage. - The initial restricted-network dependency setup failed; unchanged preflight passed with normal network access. Initial TypeScript failures in test pipe assertions were corrected. No failed attempt is counted as a pass. - Review found a missing current-master-key check. A real restart regression with all child checkpoints complete reproduced unsafe publication after switching to a different valid master key. Follow-up `7ea92c2abd9ffdb2e7dd81cecf62c8fc7a6e74a5` delegates to the canonical snapshot-key owner before both durable and in-memory cache hits. - The fixed process regressions passed 2/2. Ten key-authority and existing identity/source guard cases passed. The wrong-key process now refuses with zero additional child payload GETs, no publication, and unchanged original journal/snapshot rows. Server typecheck, module boundaries and diff checks passed on the follow-up. - Review also found that transient database failures could make the optional checkpoint cache stop a valid import. Read and save now fall back only for SQLSTATE `55P03`, `57014`, `40P01`, and `40001`. Connection loss, identity changes, crypto failures, programming errors and cancellation remain fatal. Reads use a read-only transaction with one-second lock and eight-second statement limits. - The checkpoint database-fault regression failed two cases before the fix. All four recoverable import cases then passed. The final selected run passed 21 cases, including fatal-error and cancellation controls, real row contention, both process-restart cases and key/identity guards. Server typecheck and boundaries passed. Fault injection is limited to the checkpoint database boundary; the importer, PostgreSQL, Git, archives, storage and publication remain real. - Latest master was merged without rewriting published history. Six real import/process-restart cases passed again on final head `8c8e4213a680ebb5c8d6864416bc3a4f271e1e76` in 38.03 seconds. [Required CI](https://github.com/paperclipai/paperclip/actions/runs/42658248222) finished with 54 successful checks and four intentional skips; Snyk passed. [Fresh review](https://github.com/paperclipai/paperclip/pull/20311#issuecomment-6745354381) is 5/5 with all four threads addressed and resolved. Hosted verification after normal release delivery remains pending. ## Risks - A stale receipt could skip authentication of a different object. Complete reference/lineage/namespace binding, fresh strong HEAD and row-locked save validation prevent that reuse. Missing or unsupported hints trigger full verification. - Mutable local files can change between attempts. Their source locks and fresh captures remain required; a checkpoint never certifies current local bytes. - Checkpoint persistence can contend with other snapshot owners. It has one-second lock and eight-second statement limits; the four classified transient SQL errors retain the normal fully verified path. Other database errors stop the import so lost source ownership cannot be hidden. - No schema, archive format, public API or saved-run identity change. Older snapshots without receipts remain readable and get full verification. Existing data and recovery obligations stay intact. ## Model Used OpenAI Codex, GPT-6.1 Sol (`gpt-6.1-sol`), with reasoning, repository tools and code execution. Context-window size is not exposed by the runtime. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [x] All Paperclip CI gates are green - [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge |
||
|
DottaandClaude Opus 5.5
|
099adb2728 |
fix(recovery): settle a remote workspace save that stopped without finishing (#20291)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - Native runs in remote sandboxes save the workspace as snapshots.
Each save is a `workspace_operations` row with phase
`remote_workspace_snapshot`
> - The ownership registry lists that kind as in-progress work, but it
had no settle step
> - When a restart cut off a failing final save, the row stayed
`running` after its invocations had all stopped and its run had ended
> - The retained-source fallback refuses while any workspace operation
is running, so a task whose saved files could not be verified could
never continue from its last save
> - This pull request lets the ownership reconciler settle that row as
failed
> - The benefit is that the automatic fallback can continue such a task
without any user action
## Linked Issues or Issue Description
**What happened?**
A native run in a remote sandbox finished its turn. Its final workspace
save failed because the sandbox disk was full ("No space left on device"
in the snapshot copy). A graceful restart then interrupted the retry.
The run ended as `native_workspace_sync_out_retry_exhausted`. Its
`remote_workspace_snapshot` row kept `status: running` for six days,
with `saveWork: { activeIds: [], unknown: false, quiescedAt: <time> }`
and `exportCleanup: { state: "pending" }`. The retained-source fallback
checked the task and refused with `workspace_operation_running`, so the
task stayed blocked.
**Expected behavior**
A save with no active or unknown invocation, whose run is terminal, is
finished work. The reconciler must mark it failed, so the fallback can
continue from the last committed save. A save that might still be
running must stay untouched.
**Steps to reproduce**
1. Run a native task in a remote sandbox. Make its final save fail, then
restart the server during the retry, so the save's work is quiesced and
the operation stays `running`.
2. Let the task's recovery refuse with an unverifiable-source failure
(for example `allocation_ownership_ambiguous`).
3. The retained-source fallback refuses with
`workspace_operation_running` on every check, and the task never
continues.
**Paperclip version or commit**
`master` at
|
||
|
DottaandClaude Opus 5.5
|
fd2e7482c0 |
fix(runner): stop a lost runnerd's provider and tools from a durable lineage journal (#20270)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - A local native run uses one runner daemon (`paperclip-runnerd`).
runnerd starts the provider (for example OpenCode), and the provider
starts tool processes
> - When runnerd dies in the middle of a tool call, the provider and its
tools are re-parented. The only record of that process tree was in
runnerd's memory
> - Close then cannot prove that those processes stopped. It quarantines
the session, and the next attempt stops with
`native_session_cleanup_quarantined`: "Automatic recovery stopped /
operator recovery required"
> - Restarts and runner loss are product failures. Interrupted work must
recover by itself, and it must never get two writers in one workspace
> - This pull request makes runnerd journal its process lineage to disk,
and makes close stop and prove gone every journalled process after
runnerd is lost
> - The benefit is that a killed runnerd leads to an automatic, safe
retry within the normal attempt budget, instead of an operator stop
## Linked Issues or Issue Description
Refs #20259 (runner loss classification and the lineage sampler panic).
Refs #20260 (automatic continuation of an inert cleanup quarantine).
**What happened?**
A chat eval killed only the owned local `runnerd` process (SIGKILL)
during an OpenCode tool call
(`chat-features.runner-opencode.local.recovery.retry-exhausted`,
`recovery.explain-stall` and `recovery.source-action`). Attempt 1
recorded `native_runner_process_exited` with "Retry scheduled 1/3".
Attempt 2 stopped with `native_session_cleanup_quarantined`. The task
was blocked with no owner, and the next action was operator recovery.
**Expected behavior**
After runnerd is lost, the product stops the provider and tool processes
that runnerd started, proves them gone, and continues the run within its
retry budget. When it cannot prove that, it keeps the fail-closed
quarantine.
**Steps to reproduce**
1. Start a local native OpenCode (or Codex) run.
2. Let the agent run a tool that leaves the provider's session (for
example `nohup job &`).
3. Send SIGKILL to the `runnerd` PID only.
4. Close rejects with `runner did not durably suspend before
checkpoint`, and the next attempt fails with
`native_session_cleanup_quarantined`.
**Paperclip version or commit**
master at
|
||
|
Dotta
|
5c73633a8f |
fix(workspaces): preserve remote recovery holds during online import (#20298)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - Online workspace import preserves old controller-local files while unrelated remote work continues. > - A task can have an old local checkout and a separate durable remote workspace. > - A terminal remote run can retain a file-recovery hold on that task. > - The legacy source checker rejects remote references, so that hold blocks independent local imports. > - This pull request proves that the hold belongs only to the separate remote source and fences its finalizer during import. > - Old files can migrate while the remote recovery obligation stays intact. ## Linked Issues or Issue Description Refs #19873 and #19854 for the existing cohort and source-verification paths. Searched related online migration and remote recovery PRs; no duplicate fix was found. **What happened?** Online import rejects a legacy Git checkout with `online_workspace_source_scope_unproven` when a same-task failed run has a prepared reference to a separate durable remote workspace. The legacy run reader rejects remote references before the hold can be classified. **Expected behavior** Independent local files can import under their source fences. The separate remote workspace, dirty writer, allocation, accepted result and recovery hold must stay intact. **Steps to reproduce** 1. Create related local Git workspaces. 2. Record a terminal failed native run and an exhausted file-recovery hold for a separate durable remote workspace on the same task. 3. Import the local leaf online. The old checker refuses before capture. **Paperclip version or commit** Reproduced on master at `052c638b519ebb15f81e4fcd1a96adf6cd3b2f10`. **Deployment mode** Managed server with PostgreSQL and remote Runner execution. ## What Changed - Add a distinct remote-only preservation proof. Keep remote runs excluded from legacy export sources. - Use the canonical source inspector and prepared-reference matcher. Bind the run, finalizer, action, task, accepted result, authority, allocation and prior committed snapshot. - Acquire the held run's finalizer fence through the existing sorted lock path. Existing source-proof reinspection rejects changes before publication. - Preserve previous-release prepared references with an older baseline snapshot. Do not require that snapshot's epoch or source run to equal the current run. - Add real PostgreSQL import, authenticated cold restore, preservation, refusal, lock-contention and publication-race regressions. Document the behavior. ## Verification - Genuine baseline failure on unchanged product: the real importer refuses with `online_workspace_source_scope_unproven`. - Final focused regression file: 23 tests passed in 13.19 seconds through the canonical isolated PostgreSQL wrapper on Node 24.22.2. - 24 existing affected migration/recovery tests passed against the same final production bytes. Another 83 tests were explicitly excluded by the focused selection; no full-suite pass is claimed. - The positive imports local bytes, confirms the remote finalizer fence is held during upload, leaves all remote records unchanged, parks the original repository and authenticates/restores the archive. - Refusal and race cases cover malformed, foreign, same-local and mixed references; active owners/writers; changed allocation, result, task, action, authority and baseline; and finalizer contention. - Server TypeScript check, module-boundary check and `git diff --check` passed. - An intermediate run had one incorrect assertion that expected an explicit empty legacy-source array. The canonical journal omits that empty field. The final assertion accepts absence and still forbids any exported remote source. Final relevant results are 23 new plus 24 existing passes, rather than a claim that the earlier invocation passed in full. - Required CI passed on head `6a278e717db2b5f614b0a595f99bfd0fb5430227`: 55 successful checks and 4 intentional skips. [CI run](https://github.com/paperclipai/paperclip/actions/runs/42650025033). Greptile reviewed that head at 5/5 with no actionable findings or open inline/review comments. Hosted continuation remains pending normal delivery. ## Risks - A false disjointness proof could admit an unsafe local import. Strict company/task/source binding, the canonical inspector, finalizer fencing and publication revalidation retain refusal on uncertainty. - This does not recover or settle remote files. Dirty authority and the original recovery obligation remain unchanged. - No schema, public API, archive format or saved-run identity changes. Existing persisted state carries forward. ## Model Used OpenAI Codex, GPT-6.1 Sol (`gpt-6.1-sol`), with reasoning, repository tools and code execution. The runtime does not expose the context-window size. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [x] All Paperclip CI gates are green - [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge |
||
|
Nicky LeachandPaperclip
|
052c638b51 |
fix(environments): honor stopped-task proof after warm preparation (#20289)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - A warm plugin Runner retains its task for later runs. > - A follow-up can find that the retained task has stopped. > - The resume path already asks settle to check status before it sends stop. > - A saved startup marker disabled this order, even after warm preparation had finished. > - This pull request applies the same stop-proof check to all observed-stop releases. > - The next run skips a repeated stop when status proves both execution and required preparation have stopped. ## Linked Issues or Issue Description Fixes: #20224 Refs #20203 ## What Changed - Remove the startup-marker exclusion from the existing status-first path. Keep the existing shared confirmation check. It still requires `preparationStopped: true` for startup work. - Keep stop followed by status when the first status lacks the required proof. Keep the startup drain, credential writeback, complete, and cleanup failure handling. - Add retained-task coverage with the startup metadata that warm preparation saves. Cover a startup cleanup whose first status lacks preparation proof. - Update the plugin task lifecycle contract. ## Verification - Local macOS, source base `3e8744fa91`: from `server`, `pnpm exec vitest run src/__tests__/environment-task-runtime.test.ts src/__tests__/plugin-warm-tasks.integration.test.ts src/__tests__/plugin-task-credential-writeback.test.ts` passed: 3 files, 82 tests. The integration tests use real embedded PostgreSQL and controlled plugin results. - After restoring the fix, the two affected files passed again: 73 tests. - Independent parent verification on commit `19f720ff68`: the two files filtered to the changed startup-proof and retained-task cases passed (4 tests). Independent source review found no actionable issue. - Baseline proof: with only the source condition restored to upstream, the new retained-task test with saved startup metadata fails because it sees an extra `stop`. The case without that metadata passes. The source fix was then restored. - The retained-task regression asserts exactly `environmentResumeLease`, `status`, `complete`, and `environmentAcquireLease`, with both task operations bound to the old task. The startup test also proves that missing preparation proof sends `stop` and a second `status`. - Test setup: `node scripts/agent-preflight.mjs --skip-rust-debug --explain` installed dependencies, then stopped because Cargo is unavailable on this Mac. Initial test collection failed because the plugin SDK output was absent. `pnpm --filter @paperclipai/plugin-sdk build` built that required dependency and the shared package; the targeted test run then passed. - Standard Greptile reviewed commit `19f720ff686058b4dd6b34f27cdfb9c8ff60e50a` at 5/5 with no review threads. [Review](https://github.com/paperclipai/paperclip/pull/20289#issuecomment-6743829614). - Full local checks were not run, as requested. [GitHub PR workflow 42644023868](https://github.com/paperclipai/paperclip/actions/runs/42644023868) passed on final commit `19f720ff686058b4dd6b34f27cdfb9c8ff60e50a`. It supplied broad typecheck, build, and test coverage. The separate Snyk status passed. Optional Storybook checks were skipped. - The change removes one provider stop call on the reported path. The issue's measured eight calls would become seven if other calls stay the same. The earlier five-call target is not claimed. No deployed latency measurement is claimed. ## Risks - A provider that cannot prove preparation stopped still gets stop and status. That fallback can cost one extra status call, as the existing status-first contract already allows. - A status failure keeps cleanup pending and blocks replacement. No termination proof or credential handoff is skipped. - No schema, API, dependency, or timer change. No added work on successful task reuse or ordinary task dispatch. - Live provider acceptance remains unrun. The local evidence checks orchestration and stop-proof handling. ## Model Used - OpenAI Codex, GPT-6 family, with reasoning, repository inspection, shell tools, and code execution. The exact serving model ID and context window are not exposed by this session. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [x] All Paperclip CI gates are green - [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge Co-authored-by: Paperclip <noreply@paperclip.ing> |
||
|
DottaandClaude Opus 5.5
|
dc9ad17515 |
perf(storage): time workspace snapshot save and restore steps and count storage retries (#20280)
## Thinking Path
- Cloud durable storage is a major source of failures and slowness on
staging. For example, a staging task failed when saving files after
automatic retries.
- Before measuring or fixing it, we need to see where storage time goes.
Today a saved-workspace restore before the turn appears only inside the
coarse `workspace_environment_realize` phase.
- After the turn, the save's encrypt, put, verify and multipart throttle
backoff have no timing at all. Only `captureMs`, `transferMs`,
`verifyMs`, `uploadMs` and `publishMs` exist, in `saveTiming`.
- Storage retries happen in the SDK and in the multipart uploader,
invisibly.
- This adds the measurement first, so the storage-mode chaos waves
(MinIO behind a fault proxy) can report save and restore p50/p95 and
point at the dominant wait.
## Linked Issues or Issue Description
**What happened?**
Durable-storage saves and restores were opaque in the run log and save
timing. A slow or repeatedly retried storage call could not be told
apart from slow capture, encryption or sandbox work.
**Steps to reproduce**
1. Run a remote task with S3-compatible durable storage, so it saves
after the turn and restores on the next run.
2. Inspect the run log and the `remote_workspace_snapshot` operation's
`saveTiming`. The restore's download, decrypt and verify, and the save's
put, verify and throttle backoff, have no timing.
**Expected behavior**
Each storage step has a duration, and storage request retries and
backoff are counted.
## What Changed
- `server/src/storage/request-stats.ts` (new): an async-local collector
of storage request statistics. Counts only:
- `requests`, `attempts` and `failedRequests`;
- `retryDelayMs` (SDK waits);
- `parts`, `throttled` and `backoffMs` (the multipart uploader's waits).
- `server/src/storage/s3-failure.ts`: `withS3FailureDiagnostics`, which
wraps every S3 send, records each request's SDK `$metadata.attempts` and
`totalRetryDelay`, on success and on failure.
- `server/src/storage/s3-multipart.ts`: records each uploaded part and
each resend wait, flagged throttled when the backend returned 503, 429
or SlowDown.
- `server/src/services/remote-workspace-snapshot-storage.ts`:
- Each step is a `storage.snapshot.*` run phase:
- save: `prior_attempts`, `encrypt`, `digest`, `put`, `verify`;
- restore: `download`, `decrypt`, `verify_plaintext`, `publish`.
- A caller can pass a `timing` record that receives the step durations,
the bytes and the request statistics.
- `server/src/services/remote-workspaces.ts`:
- The save passes a timing record, kept in `saveTiming.storage`.
- The sandbox unpack during a restore is its own phase,
`workspace.snapshot.sandbox_restore`.
- `execution-preparation-timing.ts`: the new phase names.
- `doc/run-log-events.md`: a new "Durable storage timing" section.
Because the restore runs inside run preparation, its steps of 50 ms or
more become `run.performance.span` rows with no extra plumbing. This is
run-log and instance-database data only; there's no Telemetry change.
## Verification
- `server`: `storage-s3-multipart-backoff`, `storage-s3-multipart`,
`storage-s3-multipart-throttle`, `storage-s3-failure-diagnostics`,
`remote-workspace-snapshot-storage` and `execution-preparation-timing`
tests: 86 passed. They include:
- collected stats for a throttled multipart upload: 12 requests, 7
failed, 7 throttled, 35.5 s backoff and 3 parts;
- a save and restore that fill every step duration and the byte count.
- `remote-workspaces.integration.test.ts`: 112 passed. Its `saveTiming`
assertion now includes `storage`.
- Server typecheck: no errors in the touched files.
## Risks
- **Overhead:** one async-local lookup per S3 request, plus timestamps
per snapshot step.
- **Collector scope:** the collector only exists inside a snapshot save
or restore. Other storage callers are unaffected; they record nothing.
- **`saveTiming`** gains a `storage` object. Readers that compare it
exactly (one test, updated) will see the new field.
## Model Used
Claude Opus 5.5 (`claude-opus-5-5`, Anthropic), 1M context window, tool
use and code execution, via Claude Code.
## Checklist
- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have searched GitHub for duplicate or related PRs and linked
them above
- [x] I have either (a) linked existing issues with `Fixes: #` / `Closes
#` / `Refs #` OR (b) described the issue in-PR following the relevant
issue template
- [x] I have not referenced internal/instance-local Paperclip issues or
links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip`
URLs)
- [x] My branch name describes the change (e.g. `docs/...`, `fix/...`)
and contains no internal Paperclip ticket id or instance-derived details
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [ ] All Paperclip CI gates are green
- [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [x] I will address all Greptile and reviewer comments before
requesting merge
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
||
|
DottaandClaude Opus 5.5
|
ac0d0374f2 |
perf(issue-detail): measure task opens on real instances (#20276)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - People open a task from their issues list to see what an agent is doing and to reply > - On a busy staging instance, the composer and the latest messages take 1.1–2.0 s to appear after a click. For one task with a slow run log, they took 13.9 s > - Local fixtures do not reproduce this. They do not have real network round trips, real server contention, or real object-store latency > - This pull request adds the measurement we need: a browser harness that times a task open on a real instance, and a Server-Timing entry on every API response that separates server time from network time > - It also records the staging baseline and the improvement plan > - The benefit is that each later performance change can be measured on real data with the same tool, before and after deploy ## Linked Issues or Issue Description **Problem or motivation** Opening a task from the issues list is slow on instances with real data. The composer and the latest chat messages can take seconds to appear. We could not measure this on staging in a repeatable way. We also could not tell server time from network and queueing time for most API requests. **Proposed solution** Add a reusable browser harness for task opens, and add a `paperclip_api` Server-Timing entry to every API response. **Alternatives considered** Local Playwright fixtures only. They do not reproduce real network round trips, server contention or object-store latency, so they did not show the delays that people see. Per-endpoint Server-Timing only. It does not cover the many other requests that run during a task open. **Roadmap alignment** This is measurement for existing task-loading behaviour. It does not add product features. The staging baseline and the measured causes are in `doc/plans/2026-11-09-task-loading-performance.md`. Follow-up pull requests change the loading behaviour itself, starting with #20277. ## What Changed - Added `server/src/middleware/api-server-timing.ts`. Every `/api` response now has a `paperclip_api;dur=<ms>` Server-Timing entry. The value is the server time from request arrival until headers are written, including admission waits. The entry is appended, so the existing `paperclip_issue` phases stay first. - Registered the middleware first in `server/src/app.ts`. - Added `tests/perf/issue-detail/staging-click-harness.js`. It clicks a task row and records, frame by frame, when the composer and the latest message become visible, layout shifts, movement of the latest message, and the requests that run before the page is ready. The summary labels each sample instead of using the task reference. It replaces query values and every path segment that contains a digit. A fetch that is still running when the sample ends is listed as pending. - Documented the harness in `tests/perf/issue-detail/README.md`. - Added the dated plan `doc/plans/2026-11-09-task-loading-performance.md` with the staging baseline, the causes, and the prioritized slices. ## Verification - `cd server && npx vitest run src/__tests__/api-server-timing.test.ts`. 3 tests pass. They cover the JSON responses, the time spent in earlier middleware, preserving route phases, and explicit `writeHead` calls. - `tests/perf/issue-detail/issue-detail.perf.spec.ts` reads the first `dur=` value. That value is still `paperclip_issue`. - The harness was used on staging to capture the baseline in the plan document. ## Risks - Low risk. The header has only a duration, with no request, user or company data. - The middleware wraps `res.writeHead` once for each request. This is the same pattern that `on-headers` uses. - Server-Timing is visible to same-origin scripts only. No `Timing-Allow-Origin` header is added. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, used in Claude Code with tool use (shell, browser automation, file editing). ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [ ] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
1004e55180 |
fix(recovery): stop stall wakes from failing setup on a held workspace (#20273)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Recovery tells a waiting parent or dependent when a child or blocker stalls on the board (`issue_blocker_stalled`) > - A legacy child that an older release bound to its parent's remote workspace failed its save, and its save recovery was exhausted. Its ended run still holds the shared workspace, days past the one-hour holder cutoff > - Workspace reservation refuses a run on such a hold and fails setup with "The workspace still has an unfinished writer or save operation." Every stall wake of the parent created a run that could only fail this way > - The stall sweep treated each failed run as an undelivered wake and retried it every ten minutes, with no backoff and no limit. On staging this made 72 failed runs (58 of 60 failures since a deploy), in pairs every ten minutes > - This pull request makes the stall sweep hold a wake while the waiter's workspace hold refuses a run, and bounds retries of a wake whose run fails > - The benefit is that a held workspace no longer produces a failed run every ten minutes, and the wake still arrives once the hold settles ## Linked Issues or Issue Description **What happened?** A parent task was woken by `issue_blocker_stalled` twice every ten minutes. Each run failed in setup (`setup_failed`, `executionRecovery.kind=bootstrap`, `providerWorkStarted=false`) with "The workspace still has an unfinished writer or save operation." The parent's remote workspace was held by a child's ended writer run whose save recovery was exhausted (`native_workspace_sync_out_retry_exhausted`, last failure `agent_stop_unconfirmed`). The child shared the parent's workspace because an older release created it before subtasks got their own workspace (#19860). **Expected behavior** A wake that cannot start because its workspace is held must not create a failed run every time. It should wait until the hold settles. Automatic retries must be bounded. **Root cause** - `server/src/services/heartbeat/workspace-preparation.ts` rethrows `RemoteWorkspaceWriterActiveError` as a setup failure when the ended writer is silent past `WORKSPACE_BUSY_HOLDER_STALE_AFTER_MS` (documented, intended). - `server/src/services/recovery/blocker-stall-wakeups.ts` retried every non-delivered wake row (including `failed`) after `REFUSED_WAKE_RETRY_MS` (10 minutes) with no backoff and no limit, and it did not check whether the waiter could start at all. ## What Changed - `remote-workspace-writer-hold.ts`: add `remoteWorkspaceHoldRefusesStart`, the one rule for a hold that reservation refuses outright (unattributed writer, or a writer silent past the cutoff with no live owner fence). - `workspace-preparation.ts`: use that rule instead of an inline copy, so reservation and wake senders cannot drift. - `blocker-stall-wakeups.ts`: before waking a waiter, check for proof that its run would be refused. The proof is that the assignee's own latest run on the task failed setup with the unfinished-writer error, and the bound workspace hold still refuses a run. Then do not wake it; count it as deferred and re-check at most hourly. A different assignee (for example a controller-only agent that never uses the workspace), a later run, or a settled hold lets the wake through. Reservation already tried and failed the unchanged-writer release for that run, so the sweep does not repeat that proof. - `blocker-stall-wakeups.ts`: count failed runs that carry the stall (the keyed wake's run and keyless resource-wait retries that keep its context). Retry after 10, then 20 minutes, and stop after the third failed run. Refusals (wake-on-demand off, agent paused) keep the 10-minute retry. A new stall (new recovery action) wakes again. - `remote-workspace-writer-hold.ts` / `remote-workspaces.ts`: share the unfinished-writer error text as one constant. - `doc/execution-semantics.md`: document both rules. ## Verification - Red before the fix (`server/src/__tests__/issue-blocker-stall-wakeups.test.ts`): 3 new tests failed. The held-workspace test got a delivered wake; the backoff test got a third wake 11 minutes after the second failure. - Green after the fix: `npx vitest run src/__tests__/issue-blocker-stall-wakeups.test.ts` passes 22/22. Added after review: a waiter with no refused run, or a reassigned waiter, is still woken; failed resource-wait retries count toward the limit. - The new tests use the staging shape: a legacy `shared_workspace` remote workspace owned by the parent, a child bound to it, the child's failed native writer run ended four days ago with `native_workspace_sync_out_retry_exhausted`, a dirty authority, and the child's board-owned exhausted recovery action. The parent's own earlier run failed setup on the hold, as on staging. They also prove that a writer inside the cutoff still gets its wake (the run parks in a workspace wait), and that the wake is delivered once the hold settles. - Related suites pass: `heartbeat-remote-workspace-busy.test.ts`, `heartbeat-dependency-scheduling.test.ts`, `native-cancelled-workspace-preservation.test.ts`, `src/modules/run-dispatch`. - `npx tsc --noEmit -p server` passes. - `pnpm verify` exits 2 because `doc/execution-semantics.md` is a global CI input; required CI supplies the broad coverage. ## Risks - Low. No schema or persisted-shape change. Saved state from the previous release (existing wake rows, existing workspace authority) is read as-is: earlier `failed` stall wake rows count toward the new three-run limit, which stops the current loop at once after deploy. - A parent whose workspace stays held is not woken for the stall until the hold settles. The stall is still listed on its page and the child's recovery action stays with the board. - Not fixed here: the exhausted save hold itself is never settled automatically (`agent_stop_unconfirmed` exhaustion in workspace recovery), and legacy children that still share a parent's workspace are not moved. Both touch workspace recovery code that is under concurrent change and are left for a follow-up. > For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and discuss it in `#dev` before opening the PR. Feature PRs that overlap with planned core work may need to be redirected — check the roadmap first. See `CONTRIBUTING.md`. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, used through Claude Code with tool use (shell, file edits, browser read-only inspection of staging). ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
2dd7a98753 |
fix(recovery): return a settled restore hold's warm sandbox to reuse (#20278)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Recovery tells a waiting parent or dependent when a child or blocker stalls on the board (`issue_blocker_stalled`) > - A legacy child that an older release bound to its parent's remote workspace failed its save, and its save recovery was exhausted. Its ended run still holds the shared workspace, days past the one-hour holder cutoff > - Workspace reservation refuses a run on such a hold and fails setup with "The workspace still has an unfinished writer or save operation." Every stall wake of the parent created a run that could only fail this way > - The stall sweep treated each failed run as an undelivered wake and retried it every ten minutes, with no backoff and no limit. On staging this made 72 failed runs (58 of 60 failures since a deploy), in pairs every ten minutes > - This pull request makes the stall sweep hold a wake while the waiter's workspace hold refuses a run, and bounds retries of a wake whose run fails > - The benefit is that a held workspace no longer produces a failed run every ten minutes, and the wake still arrives once the hold settles ## Linked Issues or Issue Description **What happened?** A parent task was woken by `issue_blocker_stalled` twice every ten minutes. Each run failed in setup (`setup_failed`, `executionRecovery.kind=bootstrap`, `providerWorkStarted=false`) with "The workspace still has an unfinished writer or save operation." The parent's remote workspace was held by a child's ended writer run whose save recovery was exhausted (`native_workspace_sync_out_retry_exhausted`, last failure `agent_stop_unconfirmed`). The child shared the parent's workspace because an older release created it before subtasks got their own workspace (#19860). **Expected behavior** A wake that cannot start because its workspace is held must not create a failed run every time. It should wait until the hold settles. Automatic retries must be bounded. **Root cause** - `server/src/services/heartbeat/workspace-preparation.ts` rethrows `RemoteWorkspaceWriterActiveError` as a setup failure when the ended writer is silent past `WORKSPACE_BUSY_HOLDER_STALE_AFTER_MS` (documented, intended). - `server/src/services/recovery/blocker-stall-wakeups.ts` retried every non-delivered wake row (including `failed`) after `REFUSED_WAKE_RETRY_MS` (10 minutes) with no backoff and no limit, and it did not check whether the waiter could start at all. ## What Changed - `remote-workspace-writer-hold.ts`: add `remoteWorkspaceHoldRefusesStart`, the one rule for a hold that reservation refuses outright (unattributed writer, or a writer silent past the cutoff with no live owner fence). - `workspace-preparation.ts`: use that rule instead of an inline copy, so reservation and wake senders cannot drift. - `blocker-stall-wakeups.ts`: before waking a waiter, read its bound remote workspace hold. When the hold refuses a run, do not wake it; count it as deferred and re-check at most hourly. A workspace with an unchanged-release baseline is left to reservation, which can release it. - `blocker-stall-wakeups.ts`: a stall wake whose run failed is retried after 10, then 20 minutes, and not after the third failed run. Refusals (wake-on-demand off, agent paused) keep the 10-minute retry. A new stall (new recovery action) wakes again. - `doc/execution-semantics.md`: document both rules. ## Verification - Red before the fix (`server/src/__tests__/issue-blocker-stall-wakeups.test.ts`): 3 new tests failed. The held-workspace test got a delivered wake; the backoff test got a third wake 11 minutes after the second failure. - Green after the fix: `npx vitest run src/__tests__/issue-blocker-stall-wakeups.test.ts` passes 20/20. - The new tests use the staging shape: a legacy `shared_workspace` remote workspace owned by the parent, a child bound to it, the child's failed native writer run ended four days ago with `native_workspace_sync_out_retry_exhausted`, a dirty authority, and the child's board-owned exhausted recovery action. They also prove that a writer inside the cutoff still gets its wake (the run parks in a workspace wait), and that the wake is delivered once the hold settles. - Related suites pass: `heartbeat-remote-workspace-busy.test.ts`, `heartbeat-dependency-scheduling.test.ts`, `native-cancelled-workspace-preservation.test.ts`, `src/modules/run-dispatch`. - `npx tsc --noEmit -p server` passes. - `pnpm verify` exits 2 because `doc/execution-semantics.md` is a global CI input; required CI supplies the broad coverage. ## Risks - Low. No schema or persisted-shape change. Saved state from the previous release (existing wake rows, existing workspace authority) is read as-is: earlier `failed` stall wake rows count toward the new three-run limit, which stops the current loop at once after deploy. - A parent whose workspace stays held is not woken for the stall until the hold settles. The stall is still listed on its page and the child's recovery action stays with the board. - Not fixed here: the exhausted save hold itself is never settled automatically (`agent_stop_unconfirmed` exhaustion in workspace recovery), and legacy children that still share a parent's workspace are not moved. Both touch workspace recovery code that is under concurrent change and are left for a follow-up. > For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and discuss it in `#dev` before opening the PR. Feature PRs that overlap with planned core work may need to be redirected — check the roadmap first. See `CONTRIBUTING.md`. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, used through Claude Code with tool use (shell, file edits, browser read-only inspection of staging). ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
1b09920a64 |
fix(issues): refuse agent status changes that detach another run's live execution (#20271)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - A task that an agent run executes holds a checkout lock and an execution lock (`checkoutRunId`, `executionRunId`) > - A status change away from `in_progress` clears these locks. It does not stop the run that holds them > - Agents cannot stop another agent's run: run cancel and the task-tree pause are board-only > - But an agent that can edit a task (for example a lead with the manage-active-checkouts grant over its report) can change its status. The run then keeps running and is no longer bound to its task > - In chat evals, a lead "paused" a launch tree. It moved running subtasks to backlog. The runs kept running, detached > - This pull request refuses an agent's status change when a different queued or running run holds the task's lock. The denial is final, links the task, and names the person's controls > - The benefit is that agents cannot interrupt another agent's run through a status edit, and runs are not left orphaned ## Linked Issues or Issue Description Refs #20261 (chat guidance sends a task-tree pause to the person's Pause subtree control). Refs #19394 (resume restarts the work a subtree pause interrupted). **What happened** An agent changed the status of a task while another agent's run executed it. The request succeeded. The server cleared `checkoutRunId` and `executionRunId`. The run continued, but it was no longer bound to the task. In the chat evals `steering.stop-versus-pause` and `task-coordination.pause-tree`, a lead moved running subtasks to backlog to "pause" a launch. Their runs kept running, detached. **Expected behavior** An agent cannot get the effect of interrupting another agent's run. When a different live run holds the task's lock, the agent's status change is refused before any write. The refusal says it is final. It sends the person to the task and names the controls that really stop or hold the work: the run's Stop or Interrupt, or More task actions, then Pause subtree on the parent task. The lock-holding run, board users, and paths that stop the run themselves continue to work. **Steps to reproduce** 1. Create agent A and agent B, with B reporting to A (A gets `tasks:manage_active_checkouts` through the manager chain). 2. Give B an `in_progress` task. Set `checkoutRunId` and `executionRunId` to a `running` heartbeat run of B. 3. As A (agent JWT with its own run), send `PATCH /api/issues/:id` with `{ "status": "backlog" }`. 4. Before this change: `200`, the status is `backlog`, both lock fields are `null`, and B's run is still `running`. ## What Changed - `server/src/routes/issues.ts`: new `assertAgentStatusChangeKeepsLiveRunBound`. For an agent actor, it refuses a status change when `executionRunId` or `checkoutRunId` names a different run in the same company that is `queued` or `running`. It runs in `PATCH /issues/:id` after the final status and assignee are known (execution-policy transitions included) and before any run stop or write. It also runs in recovery resolution when the source status changes. - The guard applies to any status other than `in_progress`, also to the current status, because the service clears the locks for those too. - The write is conditional on the lock the guard checked. The route re-reads the lock and passes it to `svc.update` as `expectedExecutionLock`. The service compares it under the row lock and returns `409 issue_execution_lock_changed` if a checkout landed in between. - These cases stay allowed: the lock-holding run itself, board users, requests with no status, a change to `in_progress`, `cancelled` (the route cancels the run), reassignment (the route stops the run first), `done` through PATCH while a Runner goal is active (terminalization pauses the goal and stops the run; recovery resolution does not get this exception), and a lock holder that already finished. - `packages/shared/src/issue-write-denial.ts`: new denial code `issue_status_change_live_run` (409). The copy says the refusal is final, links the task as `[identifier](/prefix/issues/identifier)`, and names Stop or Interrupt and More task actions, then Pause subtree. A new optional `issuePath` context field carries the link. The UI notice renders it through the existing generic path. - `doc/SPEC-implementation.md`: documents the rule in the checkout contract section. - Paths checked with no change needed: there is no bulk issue update route; tree holds and run cancel are board-only; `POST /issues/:id/release` already lets only the assignee's checkout run release; API-catalog actions call the same PATCH route; MCP status changes already refuse with `MCP_ACTIVE_EXECUTION`. ## Verification - Red on master, green with this change: - `server/src/__tests__/issue-stale-execution-lock-routes.test.ts` (embedded Postgres, real issue service): a manager moves its report's running task to backlog. On master: `200`, status `backlog`, `checkoutRunId`/`executionRunId` changed to `null`. With the fix: `409 issue_status_change_live_run`, row unchanged, run still `running`. - `server/src/__tests__/issue-agent-mutation-ownership-routes.test.ts`: refusals for `backlog`, `todo`, `blocked`, `in_review`, `done`, a queued lock holder, and a repeated current status (red on master). Allowed cases: field edits, the lock-holding run, board, `cancelled` (run is cancelled), and finished lock holders; an allowed change carries `expectedExecutionLock`. - Same embedded-Postgres file: `svc.update` with an `expectedExecutionLock` that differs from the row returns `409 issue_execution_lock_changed` and leaves the row unchanged. - Also passed locally: `issue-closed-workspace-routes`, `issue-comment-reopen-routes`, `issue-recovery-actions`, `issue-scheduled-retry-routes`, `low-trust-red-team-routes`, `issue-update-comment-wakeup-routes`, `external-object-routes`, `packages/shared` `issue-write-denial`, `ui` `IssueWriteDenialNotice`. - `tsc --noEmit` shows no errors in the changed files. Full typecheck, build and test are left to CI. ## Risks - Behavior change for agents: a lead or CEO that used status edits to take over a report's running task now gets a 409. The supported paths (comment, cancel, reassign, or the person's Stop or Pause subtree) still work. - A queued run that holds the lock also blocks the change. This is deliberate: the queued run would otherwise start later with no task bound to it. - Agents can still cancel another agent's run by cancelling its task. That is existing, documented behavior ("task cancellation cancels the intent"), and this PR does not change it. ## Model Used - Anthropic Claude Opus 5.5 (`claude-opus-5-5`), via Claude Code with tool use (shell, file edits, test execution). ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandPaperclip
|
a3a25ce6c9 |
fix(workspaces): stop running Paperclip's worktree setup before every agent turn (#20269)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Before an agent turn starts, the harness prepares the task's git worktree > - The harness runs a repository's `scripts/provision-worktree.sh` when the project sets no `provisionCommand` and the file exists > - The harness runs it on every turn, both when it creates the worktree and when it reuses it > - In the Paperclip repository, that script writes the worktree's isolated instance config and runs `pnpm install` > - Its "install needed" check compares against every `node_modules` directory in the base checkout, including other worktrees, so it reinstalls on every turn > - This pull request renames the Paperclip script so that the harness default does not find it in this repository > - The benefit is that Paperclip agent turns no longer wait on a 6–9 second install. Other repositories keep the harness feature. ## Linked Issues or Issue Description **What existing behavior does this improve?** Time to first token for agent turns in managed worktrees of the Paperclip repository. **Subsystem affected** Execution workspace provisioning (`server/src/services/workspace-runtime.ts`) and the repository script `scripts/provision-worktree.sh`. **Current behavior** Every agent turn in a Paperclip worktree first runs `bash ./scripts/provision-worktree.sh`. In a local test, a `pnpm install --frozen-lockfile` with nothing to change took 6.1 s and 6.3 s, and a full install from a warm store took 9.0 s. The script's fingerprint skip does not prevent this, because the script also requires every base-checkout `node_modules` path (up to depth 4) to exist in the worktree. On a developer machine that list had 1,735 entries, including other worktrees under `.claude/worktrees/` and `.worktrees/` and pnpm's internal `node_modules/.pnpm/*/node_modules`. A managed worktree never contains the other worktrees' paths, so the script always decides to reinstall. **Proposed behavior** The Paperclip repository does not ship a file at `scripts/provision-worktree.sh`. The setup script becomes `scripts/setup-paperclip-worktree.sh`. People can run it by hand or set it as the project `provisionCommand` to opt in. The harness default stays for other repositories. **Reason and benefit** Agents install dependencies when they need them (`node scripts/agent-preflight.mjs`, as `AGENTS.md` says). Many turns never run code. Instance setup stays available on demand: `scripts/provision-worktree-runtime.sh` runs the setup script when a dev server starts in a worktree that has no config. Database seeding was already deferred. **Breaking changes** A project that explicitly sets `provisionCommand` to `bash ./scripts/provision-worktree.sh` for this repository will fail with "file not found". Change it to `bash ./scripts/setup-paperclip-worktree.sh` or clear it. ## What Changed - Rename `scripts/provision-worktree.sh` to `scripts/setup-paperclip-worktree.sh`. The content is unchanged. - `scripts/provision-worktree-runtime.sh` calls the new name when the worktree has no config. - Tests read the real script from its new path (`provision-worktree-self-heal`, `workspace-runtime`, `worktree-seed-server-spawn`). Server fixtures still copy it into their test repositories as `scripts/provision-worktree.sh`, so the harness default stays covered. - `doc/DEVELOPING.md` says that the Paperclip repository deliberately has no `provision-worktree.sh`, and tells people how to opt in. ## Verification - `node --test scripts/__tests__/provision-worktree-self-heal.test.mjs`: 33 passed, 1 skipped (the skip already existed). - `cd server && npx vitest run src/__tests__/workspace-runtime.test.ts src/__tests__/heartbeat-workspace-source-trust.test.ts src/__tests__/execution-workspace-reopen.test.ts`: 242 passed on the rebased head. - `cd server && npx vitest run src/__tests__/worktree-seed-server-spawn.test.ts`: 2 passed. - Manual: in a Paperclip worktree, start an agent turn. The run log does not show a `workspace_provision` operation for the script. ## Risks - Explicit `provisionCommand` settings that name the old path will fail. See "Breaking changes". - Open pull requests that edit `scripts/provision-worktree.sh` (for example #16213, #11335, #11733, #19764) will conflict on the rename. Git rename detection should carry most edits over. - An agent that runs tests in a new worktree without first running preflight will see a missing-module error. `AGENTS.md` already tells agents to run `node scripts/agent-preflight.mjs`. - The script's install check still has the bug described above. It now runs only on demand. A separate fix can address it. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, through Claude Code with tool use (shell, file edits). Standard context window. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Paperclip <noreply@paperclip.ing> |
||
|
DottaandClaude Opus 5.5
|
2b461b2f58 |
fix(native): require publication for files named under a Deliverable label (#20231)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - Native agents finish a task with a completion report, and a server
gate checks that requested files are published to the task
> - The gate finds a file request only from a creation verb, such as
"Write report.pdf"
> - Delegated briefs often list the output under a "Deliverable:" label
instead, and the gate did not know `.html` or `.css` names
> - So a worker can finish a website task with `index.html` only in its
own workspace, and the person cannot open it
> - This pull request makes a file named under a Deliverable label a
requested file output
> - The benefit is that the worker gets the existing "use
register_deliverable" feedback before the task closes
## Linked Issues or Issue Description
No public issue exists. Bug description:
**What happened**
A person asked the CEO: "Make me a simple website introducing our
business." The CEO delegated a task. Its brief said "Deliverable: - A
self-contained static website in this task's workspace: index.html plus
a stylesheet (for example styles.css)". The worker wrote `index.html`
and `styles.css` in its workspace. It did not call
`register_deliverable`. It cited `index.html` and `styles.css` as
completion evidence, and the gate accepted the report. The task was
marked done with no attachment. The CEO then told the person that the
files were "in the task workspace".
**Expected behavior**
A brief that names a file as its deliverable is a file request. The
completion gate must reject a workspace-only file and tell the worker to
publish it with `register_deliverable`.
**Steps to reproduce**
1. Call `explicitlyRequestsFileOutput` with the brief above.
2. See that it returns `false`, so `validateNativeDeliverableEvidence`
accepts the bare `index.html` evidence ref.
## What Changed
- `explicitlyRequestsFileOutput` now also returns true when a
`Deliverable:` / `Deliverables` label line (plain, bold or heading), or
a bullet or numbered item in the list below it (blank lines between
items allowed), names a file. Relative paths such as `dist/index.html`
count.
- Under the label, web page files (`.html`, `.htm`, `.css`) count with
the existing file types.
- Items that name an input ("a summary of input.csv", "a comparison of
input.csv and baseline.csv"), describe a repository change, pull
request, commit or branch, forbid the named file (`do not create
report.pdf`), name a file to edit in place (`update styles.css`) without
asking to publish it, send the file inline (`index.html and styles.css
inline in the reply`), or forbid creating files, do not count. An inline
or no-upload instruction about another output in the same item
(`index.html and a short summary in the reply`, `index.html; do not
upload it externally`) does not cancel the named file. The
repository-change check ignores file names, so `branch-report.pdf` still
counts.
- The sentence-based detector is unchanged.
- `doc/AGENT-ARTIFACTS.md` describes the label rule.
## Verification
- `cd server && npx vitest run
src/services/native-runtime/native-deliverable-feedback.test.ts
src/services/native-runtime/native-runner-file-handoff.test.ts` (197
pass), then after the review fixes
`native-deliverable-feedback.test.ts`,
`native-completion-format-feedback.test.ts` and
`chat-attachment-reuse.test.ts` (211 pass). The 13 first-round review
cases fail on the original head.
- New cases: the real delegated brief, inline labels, heading lists,
numbered lists and bold labels require publication. Unlabelled text,
inline-chat deliverables, input references, repository changes, text
after the list ends, and "Deliverable quality ..." prose do not.
- The chat eval `onboarding.first-work` exercises this path end to end
on its candidate branch.
## Risks
- Low risk. The change only adds a true result for labelled file
outputs; an inline-only or no-file label item stays false, as the
sentence detector already decided.
- A labelled brief that names a file which is not a repository change
now needs `register_deliverable` or an accessible work product. That is
the documented delivery rule.
## Model Used
- Anthropic Claude Opus 5.5 (`claude-opus-5-5`), tool use and code
execution in Claude Code.
## Checklist
- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have searched GitHub for duplicate or related PRs and linked
them above
- [x] I have either (a) linked existing issues with `Fixes: #` / `Closes
#` / `Refs #` OR (b) described the issue in-PR following the relevant
issue template
- [x] I have not referenced internal/instance-local Paperclip issues or
links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip`
URLs)
- [x] My branch name describes the change (e.g. `docs/...`, `fix/...`)
and contains no internal Paperclip ticket id or instance-derived details
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [ ] All Paperclip CI gates are green
- [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [x] I will address all Greptile and reviewer comments before
requesting merge
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
||
|
DottaandClaude Opus 5.5
|
4c15eea770 |
fix(accounting): a run's unproven accounting never holds the company (#20254)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - After a restart, each lost provider receipt leaves an accounting
debt that settles later
> - In restart-chaos G1 (seed 9802, task t02), six debts each waited the
full 10-minute closure bound, and the board's accepted Retry waited with
them
> - The last debt settled as an unpriced charge. Under the blocking
unpriced-usage policy, that charge hard-stopped the whole company, which
cancelled the Retry, and the task showed "budget"
> - Dotta's plan decision 1 says accounting is never a company-wide
gate: admission reads per-run exposure and real hard stops only
> - This pull request makes those debts settle at once when the stop is
proven, records accounting-settled charges without a company hold, and
labels the remaining accounting wait correctly
> - The benefit is that a restart no longer stalls a company's work for
10 minutes or stops it outright
## Linked Issues or Issue Description
Refs #20250 (Phase 2: a run that only inherited its predecessor's runner
identity settles at once; this PR builds on it), #19944, #19826, #19857
**What happened?**
Restart-chaos wave w-20261109T111954-s9802 (G1), task t02, master
|
||
|
DottaandClaude Opus 5.5
|
8525bb534c |
fix(chat): post a turn's answer when a card it did not create is pending (#20220)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - People talk to agents in Agent Chat. The agent's final message becomes the chat reply. > - A chat thread can hold a pending card that the board addressed to another person, for example a human-only confirmation for Riley. > - When Alex asks the agent to act on that card, the agent correctly gets `403 interaction_human_only` and writes a final answer that explains the denial. > - The shared answered rule (`conversationTurnAnswered()`, from #20179 and #20204) counted any pending human-only card in the thread as the turn's own response. So the answer was suppressed, and the chat stayed `active` with no reply. > - This pull request limits that rule to cards that the turn's own run created. Other cards still hold the task status. > - The benefit is that the person always gets the turn's answer, unless the turn itself asked a question with a card. Live completion and restart recovery stay consistent. ## Linked Issues or Issue Description Refs #20179, #20204 (the shared answered rule this PR corrects). **What happened** In an Agent Chat, the board created a pending `request_confirmation` card with `resolverPolicy: human_only`, addressed to another user (Riley). The card has no source run. Alex wrote "I approve Riley's request." The agent tried to accept the card and got `403 interaction_human_only`. It then wrote a correct final answer: the card needs Riley. The run succeeded with a `yielded` result and a `response_wake` continuation. Run completion recorded `presentationDecision: none`. No comment was created. The conversation stayed `active`, so Alex never got a reply. Cause: `conversationTurnAnswered()` reused the status-gate query `pendingNativeInteraction()`. In conversation mode, that query excludes another run's card only when it is an ordinary question or an `anyone` confirmation. A human-only card addressed to someone else is kept. The rule then treated that card as the card that "presents the turn", although this run did not create it. **Expected behavior** A successful chat turn posts its final answer exactly once, and the conversation moves to `waiting`. Only a pending question or confirmation card that the turn's own run created presents the turn in place of its prose. A card from an earlier turn, or one the board addressed to another person, can hold the task status, but it does not suppress the answer. **Steps to reproduce** 1. Open an Agent Chat as Alex. 2. As the board, create a pending `request_confirmation` card in the chat with `resolverPolicy: human_only` and `addresseeUserId` set to another person. 3. Send "I approve Riley's request." 4. Let the agent try to accept the card, get `403 interaction_human_only`, and end its turn with a final answer. 5. See that no reply appears, `presentationDecision.commentAction` is `none`, and `conversationState` stays `active`. ## What Changed - `server/src/services/native-runtime/native-governance.ts`: `conversationTurnAnswered()` now checks only pending cards whose `sourceRunId` is the turn's run. `pendingNativeInteraction()` gets an optional `createdByRun` filter for this. The status gate (`pendingNativeGovernance()`) does not change, so another person's card still holds the task. - Both callers get the fix through the one shared rule `conversationTurnPresentsAnswer()`: live run completion (`run-completion.ts`) and background presentation recovery (`recoverCommittedNativeChatResponse`). - `server/src/services/heartbeat/run-completion.test.ts`: new live cases "someone else's confirmation card" (board-created, human-only, addressed to another user, no source run) and "earlier turn's question card" (created by an earlier run of the same agent). Each case now also checks that the conversation moves to `waiting`. - `server/src/services/native-runtime/agent-chat-response-recovery.test.ts`: new restart case for the board-created card, plus a `waiting` check after recovery. - `doc/execution-semantics.md`: state that only the turn's own card presents the turn. ## Verification - Thesis: the reply is suppressed because the answered rule counts any pending human-only card in the thread, not only this run's card. If this is true, a turn with only a board-created card addressed to someone else posts no comment on master, and restricting the rule to `sourceRunId = runId` posts exactly one comment and settles the conversation to `waiting`, with the own-card cases unchanged. - Failing first: without the fix, the two new live cases failed (`expected [] to deeply equal [answer]`). The new recovery case also failed (no comment). - After the fix: - `cd server && npx vitest run src/services/heartbeat/run-completion.test.ts -t "posts a waiting conversation turn"`: 9 passed (includes the two own-card cases, which still post nothing). - `cd server && npx vitest run src/services/native-runtime/agent-chat-response-recovery.test.ts`: 62 passed. - Both suites use a real embedded PostgreSQL database. - Turn start: no change. The rule runs only after the provider turn ends. ## Risks - Low. Only the presentation rule changes. The status gate, the conversation settle rule (it already required `sourceRunId = run.id`) and the external-chat path do not change. - Behavior shift: a turn now posts its answer while another run's or the board's card is pending. Before, the answer was lost. A card that the turn itself created still presents the turn, as before. - Previous-release state: no persisted shape change. A run that the previous release already recorded as `none` keeps that decision. Recovery does not re-present it. > For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and discuss it in `#dev` before opening the PR. Feature PRs that overlap with planned core work may need to be redirected — check the roadmap first. See `CONTRIBUTING.md`. ## Model Used - Anthropic Claude Opus 5.5 (`claude-opus-5-5`), with tool use (shell, file edits, test runs) in Claude Code. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
6fd46c0025 |
perf(budgets): judge Paperclip's share of restart recovery; report the provider's session load (#20183)
## Thinking Path - Speed budgets protect the time to first visible agent activity, and restart recovery is one of the judged metrics (`native-speed-budgets.ts`, wave grader). - In the B1 concurrency-6 chaos wave (`w-20261109T045320-s9807`), Codex `restart_recovery` was p50 8.6 s and p95 11.2 s, against a budget of 6.5 s and 8.0 s. At concurrency 4 (K13) it was 3.8 s and 5.0 s. - The preparation spans on a copy of the kept database break it down: - Paperclip's share (ready → claim → prepare → runtime stage → runner launch) is a steady 1.2–2.0 s. - `provider.session.load`, the Codex app-server boot plus `thread/resume` of long saved sessions, takes 3.0–6.3 s, 65–75% of each recovery. - Resumes run concurrently, and the worktree lock and the startup reconciler are not on the path. - Judging the full latency therefore mixes our code with provider and session-size effects. Budget the part we control, so a regression in our code shows up regardless of session size or concurrency, and report the rest. ## Linked Issues or Issue Description **What happened?** `restart_recovery` breached its budget whenever many long provider sessions resumed at once, although Paperclip's own recovery work stayed within about 2 s. The breach did not point at anything Paperclip could fix, and a real regression in our code could hide inside the provider-load variance. **Steps to reproduce** 1. Run a local chaos wave at concurrency 6 with long-running Codex tasks and K1/K2/K4 faults (for example B1, seed 9807). 2. Compare `restart_recovery` in `speed.json` with `provider.session.load` in the recovered runs' run logs. **Expected behavior** The budget judges Paperclip's share of a recovery. The provider's session load and the full latency are reported but not judged. ## What Changed - `server/src/services/native-runtime/native-speed-budgets.ts`: - New budgeted metric `restart_recovery_control_plane`: from `ready` to the start of `provider.session.load`, or to the next provider event when nothing reloads (for example a reattached runner). - Its budget is 2.5 s / 4.0 s for claude, codex and opencode. Measured: 1.2–2.0 s per run at concurrency 6. - The full `restart_recovery` stays in the table for reporting. - Breach payloads carry `restartTotalMs` and `providerLoadMs`. - `server/src/services/native-runtime/native-run-trace.ts`: - The per-run check records when `provider.session.load` starts, keeps the recovery measurement, and judges it when the run finishes. - It judges after the deferred diagnostics drain, so a provider span committed after the provider event still counts. - It emits `restart_recovery_control_plane` breaches only. - `tests/restart-chaos/ttft-baseline-stats.ts`: each recovery sample gains `controlPlaneMs` and `providerLoadMs`. The provider-load start is the span event's write time minus its duration. - `tests/restart-chaos/wave-speed.ts`: - Three restart rows per adapter: `restart_recovery` (report-only), `restart_recovery_control_plane` (budgeted) and `restart_provider_load` (report-only). - All three have outage time removed. - `tests/restart-chaos/ttft-baseline.ts`: the recovery table adds a "Paperclip share" column. - Docs: `doc/run-log-events.md` (metric and payload) and the speed-budget table in `doc/plans/2026-11-07-durable-ownership-and-reconciliation.md`. ## Verification - `server`: `npx vitest run src/services/native-runtime/native-speed-budgets.test.ts`, 18 passed. Two new tests: - a recovery dominated by provider load does not breach; - a slow Paperclip share breaches with `restartTotalMs` and `providerLoadMs`, even when the provider span is committed after the provider event. The existing recovery tests moved to the new metric. All six were red before the trace change. - `tests/restart-chaos`: `npx vitest run --config tests/restart-chaos/vitest.config.ts wave-speed.test.ts ttft-baseline-stats.test.ts`, 29 passed. New tests cover the sample split and the grader rows (a judged share, a report-only provider load, zero breaches). - Server typecheck: no errors in the touched files. ## Risks - **Lower sensitivity to provider-side slowdowns.** A provider regression in session load no longer breaches. It stays visible in the report-only rows and in breach payloads. - **Runs with no `provider.session.load` span** (reattach, or an adapter that doesn't emit it) are judged on the full latency against the control-plane budget. That's conservative. - **Run-log consumers that filter on `metric: "restart_recovery"`** see `restart_recovery_control_plane` from now on. The doc notes the change date. ## Model Used Claude Opus 5.5 (`claude-opus-5-5`, Anthropic), 1M context window, tool use and code execution, via Claude Code. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
ad2304dc88 |
fix(chat): recover a held conversation turn's answer after a restart (#20204)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - People talk to agents in Agent Chat. The agent's final message becomes the chat reply. > - A chat turn can end while its task stays held: a pending approval, an execution stage or a dependency blocker keeps the status. > - #20179 made live run completion post that held turn's answer. It added `conversationTurnAnswered()`. > - Background presentation recovery still required a `conversation_turn_finished` decision and a `done` result. > - If the server restarts between the status commit and the live presentation, recovery skips the held turn. The person never gets an answer. > - Restarts are product failures. A committed answer must not be lost. > - This pull request makes recovery use the same answered rule as the live path, through one shared helper. > - The benefit is that a held turn's answer reaches the chat exactly once, with or without a restart. ## Linked Issues or Issue Description Refs #20179 (live fix; this PR closes the "Known gap" in its Risks section). **What happened** In an Agent Chat, the agent resubmitted a hire approval and ended its turn with a yielded result. The finalizer committed `governed_response_waiting` and moved the task to `in_review`. The server restarted before run completion posted the reply. Background presentation recovery found the run as a candidate, but declined it, because it accepted only `conversation_turn_finished` with a `done` result. No reply was posted. The run stayed a pending presentation candidate. **Expected behavior** After a restart, recovery posts the held turn's final answer exactly once. It never posts a second copy when the live path already presented the answer. When the turn's own question or confirmation card presents it, recovery posts nothing and records that decision, as the live path does. **Steps to reproduce** 1. Open an Agent Chat. Link a pending hire approval to the chat task. 2. Let the agent end the turn with a yielded result and a `response_wake` continuation. The status decision commits as `governed_response_waiting`. 3. Stop the server after the status commit and before run completion presents the reply. 4. Start the server. Presentation recovery runs. 5. See that the chat shows no reply. ## What Changed - `server/src/services/native-runtime/native-governance.ts`: add `conversationTurnPresentsAnswer()`. It is the one rule for whether a successful conversation turn's prose is its answer: `conversation_turn_finished`, or a committed decision with no pending card of the turn (`conversationTurnAnswered()`). - `server/src/services/heartbeat/run-completion.ts`: the live path calls the shared helper. No change in behavior. - `server/src/services/native-runtime/native-run-finalizer.ts` (`recoverCommittedNativeChatResponse`): - A current conversation turn no longer needs `conversation_turn_finished`. The turn must still be the unsettled current turn: same agent, user, session generation, `conversationState: active`, and no other execution run. A `conversation_turn_finished` turn still needs `in_progress`. - A conversation turn with a `yielded` result is accepted next to `done`. - The resolver gets `conversationTurnFinished` from the shared helper. - When the turn's own card presents it, recovery records `presentationDecision: none` (reason `committed_conversation_interaction_presented`) and settles the conversation turn, like the live path. The run then leaves the pending presentation set. Before, such a run stayed a candidate. - `server/src/services/native-runtime/native-response-presentation-query.ts`: the candidate prefilter matches the same turns: any committed reason for an active conversation turn, and `yielded` conversation results. - `server/src/services/native-runtime/agent-chat-response-recovery.test.ts`: restart regressions with the real finalizer and a real database. - `doc/execution-semantics.md`: state that recovery applies the same rule. ## Verification - Failing first. With only the new tests applied, 5 of 7 new cases failed: pending approval, revision-requested approval, execution stage (all yielded), pending approval with `done`, and the own-card case. The run was a candidate, but recovery posted nothing and did not settle it. - After the fix: `cd server && npx vitest run src/services/native-runtime/agent-chat-response-recovery.test.ts src/services/heartbeat/run-completion.test.ts src/__tests__/idle-native-finalization.test.ts src/services/presentation-ownership-read.test.ts src/services/chat-monitor-continuation.test.ts src/services/native-runtime/native-preserved-monitor-wait.test.ts src/services/native-runtime/native-response-publication-retry.test.ts src/services/heartbeat/run-lifecycle.test.ts src/__tests__/scoped-deadline-work.test.ts src/services/native-runtime/paperclip-control-plane-port.test.ts src/__tests__/heartbeat-run-summary.test.ts src/__tests__/agent-conversations.test.ts`: 12 files, 627 tests passed. - The new cases check: - Held turns: the run is a candidate, the check posts the exact final answer once, the presentation decision is recorded, the run leaves the candidate set, and a second recovery returns false with one comment. - Live already presented: no candidate, no second comment. - Own question card: no comment, `presentationDecision: none`, the card stays pending, the conversation settles to `waiting`, and the run leaves the candidate set. - Previous-release live decision `none` (`yielded_control_plane_wait`): kept, not re-presented. - `tsc --noEmit -p server`: no errors in changed files. Four unrelated errors remain in this local worktree (`esbuild` not linked in `native-harness-prestage.test.ts`, and a linked-package type mismatch in `run-dispatch/adapters/postgres.ts`). - Turn start: no change. The helper runs only in run completion (after the provider turn ends) and in background presentation recovery. Nothing new runs before `turn.start`, and no files move. The live path keeps the same queries: it short-circuits on `conversation_turn_finished`, and otherwise runs the same card read as before. ## Risks - Low to medium. Recovery can now post the answer of a conversation turn whose task is held. This matches the live path since #20179. - A card-presented turn found by recovery is now settled with `presentationDecision: none`. Before, it stayed a pending candidate and kept the presentation worker (and idle-sleep checks) busy. - A `conversation_turn_finished` turn keeps the `in_progress` check. Held turns accept any status their gate committed. The unsettled-turn checks (active conversation state, session generation, execution run) still block stale or superseded turns. - Previous-release state: no persisted shape changes. A run committed by the previous release with no saved presentation decision (the restart window) is recovered by the new code. The test fixture uses the unchanged finalizer, so it saves exactly the previous release's shape. A run whose live path already recorded a decision keeps it. A held turn that the previous release recorded as `yielded_control_plane_wait` is not re-presented, so no late duplicate or stale reply appears. - Not changed: generic stranded-work recovery in `recovery/service.ts` still skips only `conversation_turn_finished` runs during the publication window. > For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and discuss it in `#dev` before opening the PR. Feature PRs that overlap with planned core work may need to be redirected — check the roadmap first. See `CONTRIBUTING.md`. ## Model Used - Anthropic Claude Opus 5.5 (`claude-opus-5-5`), with tool use (shell, file edits, test runs) in Claude Code. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Nicky LeachandClaude Opus 5.5
|
990de1e0bf |
perf(environments): release a stopped plugin task once (#20203)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Plugin environment drivers with `supportsTasks` run agents as provider tasks, and warm runs keep a parked task for the next run > - A follow-up run can find that its parked task already stopped. Core then retires that task before it acquires a new one > - Core sent `stop`, `status`, and `complete` in the settle step. Then the plugin release hook sent `stop`, `status`, and `complete` again for the same task. Every call was on the critical path of the new run > - This pull request removes the repeated calls and keeps every call that does work no other call does > - The benefit is a faster start for a run whose parked task already stopped ## Linked Issues or Issue Description No issue exists. This follows the enhancement template. **What existing behavior does this improve?** The release of a plugin task lease that has already stopped. This occurs when a follow-up run finds that its parked warm task stopped. **Subsystem affected** Server: plugin environment drivers and the plugin Runner task lifecycle. **Current behavior** Core sends `stop`, `status`, and `complete` in the settle step. Then the plugin release hook sends `stop`, `status`, and `complete` again for the same task. Core also sends `stop` after resume already saw that the task stopped. On a hosted test deployment, 8 of 40 follow-up runs took this path. From the resume to the new acquire, these runs spent P50 1014 ms and P90 1080 ms. These 8 runs are the full overall P90 (1275 ms). **Proposed behavior** Core does not call the release hook when settle confirmed the stop and completed the same task. After a resume with no lease, Core sends `status` first, and sends `stop` only when `status` does not confirm the stop. **Reason and benefit** The repeated calls do no new work. Removing them saves about 355 ms, plus the `stop` share of about 187 ms, for each such run. These numbers are modeled from trace timings. **Breaking changes** None for the plugin API. A task driver must release task-scoped resources in its `status` or `complete` handling. The doc now says this. **Additional context** The settle `status` stays. Providers release task storage there, and it records the stop that gates the credential writeback and `complete`. ## What Changed - `settleDispatchedEnvironmentTask` now returns `"completed"` when it confirmed the stop and completed the task in this call, `"stopped"` when the lease was already settled, and `false` otherwise. - The plugin driver release does not call `environmentReleaseLease` when settle returned `"completed"`. In all other cases it calls the hook as before. - When `environmentResumeLease` returns no provider lease, the release sends `status` first. It sends `stop` (and a second `status`) only when that `status` does not confirm `executionStopped: true`. Startup and startup-cleanup leases keep the old `stop` first order. - The cleanup sweep (`retryPendingSandboxTeardown`) and `destroyRunLease` do not change. A settle failure still leaves the lease in `pending_cleanup`, and the sweep runs `stop`, `status`, `complete`, and the release hook. - `doc/plugins/PLUGIN_RUNNER_EXECUTION.md` tells drivers about the new order and that task-scoped resources must be released in `status` or `complete`. ### Dispatch count for a stopped parked task (resume to new acquire) | | Calls | | --- | --- | | Before | resume, stop, status, (credentials), complete, release hook (stop, status, complete), acquire: 9 provider calls (10 with Codex writeback) | | After | resume, status, (credentials), complete, acquire: 4 provider calls (5 with Codex writeback) | The modeled saving is about 540 ms for each such run: about 355 ms for the repeated release, and about 187 ms for the settle `stop` and `status`. This pull request keeps the settle `status`, so the real saving is about 355 ms plus the `stop` share of the 187 ms. These numbers are modeled from trace timings. They are not measured on this branch. ### What covers each removed call - Release hook `stop`, `status`, `complete`: settle already sent `stop` and `status` and checked `executionStopped: true` for the same task id (`lease.providerLeaseId`), then sent `complete`. The skip uses settle's success as its condition, not the code path. - Settle `stop` after a resume with no lease: the next `status` must report `executionStopped: true`. If it does not, Core sends `stop` and `status` as before. ### What each kept call protects - Settle `status`: providers release task storage and cancel preparations here. It also writes `runnerTaskExecutionStopped`, which the credential writeback and the `complete` bookkeeping need. - Credential writeback: Codex can rotate its token. The writeback must land before the new task installs credentials. It still runs inside `complete`, before the new acquire. - Settle `complete`: it retires every run lease that owned the task and clears the dispatch marker. - Release hook when settle did not complete a dispatched task in this call: the provider may still hold the lease. ## Verification - `pnpm --filter @paperclipai/shared --filter @paperclipai/plugin-sdk build` - `pnpm --filter @paperclipai/server typecheck` - `npx vitest run src/__tests__/plugin-warm-tasks.integration.test.ts src/__tests__/plugin-task-credential-writeback.test.ts src/__tests__/environment-task-runtime.test.ts` - Related suites: `environment-runtime`, `plugin-runner-task.integration`, `task-creation-recovery.integration`, `heartbeat-plugin-environment`, `native-startup-cleanup-recovery`, `plugin-environment-driver-ready-recovery`, `environment-task-authority.integration`, `native-workspace-warm-reuse`, `remote-workspaces.integration`, `legacy-workspace-restore-recovery`, `plugin-task-source-recovery.integration`, `managed-plugin-environments`. Locally, all of these pass except `heartbeat-plugin-environment` "inherits the instance default environment across companies…". That test also fails locally with the `origin/master` service files, so this change does not cause it. The existing `plugin-runner-task.integration` same-run replacement test now expects no release hook after a completed settle. New tests: - The stopped parked path sends exactly resume, `status`, `complete`, acquire. The `status` goes to the old task. - A resume with no lease for a task that still runs sends `status`, `stop`, `status`, `complete`. - A failure at `status`, `complete`, or `stop` leaves the lease in `pending_cleanup`, and the sweep then sends `stop`, `status`, `complete`, and the release hook. - With an observed stop, the Codex writeback runs before `complete`. A writeback failure leaves the lease for cleanup, and the retry writes back and completes. - A startup-cleanup lease still sends `stop` first. ## Risks - A third-party task driver that releases resources only in `onEnvironmentReleaseLease` would not get that call after a completed settle. The doc now states the requirement. The cleanup sweep still calls the hook after any failure. - A driver that returns no lease from resume while the task still runs costs one extra `status`. The behavior stays correct. - No migration. No API change for plugins. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), via Claude Code with tool use and code execution. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Devin FoleyandPaperclip
|
9b71d03215 |
fix: recover interrupted native preparations and unblock upgrades (#20177)
## Thinking Path > - Paperclip manages AI-agent companies and their work. > - A native run prepares its workspace before a runner PID or checkpoint exists. > - A server crash can interrupt that stage and leave its controller on another host. > - Recovery needs a durable owner proof, a delivered recovery wake, and the saved execution contract. > - The old paths missed controller lease registration and boot proof, skipped checkpoint-free preparations, and counted retained work as live upgrade blockers. > - This PR connects preparation to the existing controller lease and recovery paths, and separates retained work from the upgrade census. > - The result is bounded recovery without a second executor or loss of saved work. ## Linked Issues or Issue Description **What happened?** A native preparation interrupted by a server crash could remain running indefinitely with no runner PID or checkpoint. Native controller stamps did not wake the shared instance lease. That lease was disabled on volume-backed Railway services, and native takeover omitted the boot binding needed for its cross-host proof. Scoped recovery selected checkpoint-backed runs but skipped these preparations. The upgrade probe then counted the retained row as current execution and could block every later upgrade. **Expected behavior** Track the preparation's controller before a runner starts. Recover it through the normal classifier after proving its owner stopped or fencing that boot. Keep live or unprovable owners held. Deliver recovery wakes to preparations without checkpoints. Preserve unresolved historical work and report it separately from current execution during upgrade observations. **Steps to reproduce** 1. Save an observed native preparation with no runner process, result, provider-start evidence or checkpoint. 2. Replace its controller with a process in another PID space. Let its owner lease expire and send a scoped recovery wake. 3. Before this fix, recovery skips the checkpoint-free preparation or cannot prove its foreign controller stopped. Upgrade readiness continues to count that old row as a live blocker. Related work: #18037, #19651, #19868, and #20168. The foreign-PID guard from #19868 remains. The queued first-turn input fix from #20168 is included through the current base. ## What Changed - Register the existing instance lease when native execution and restart claims are written. Keep passive identity reads free of ownership effects. Keep that lease enabled on volume-backed Railway services unless explicitly disabled. - Preserve the existing PID/start-time proof within the same readable PID space. Bind foreign or unknown-space takeover to its saved boot and last write. A live lease or grace period keeps the run held with a scheduled check. Repeated evidence waits preserve their timestamp. - Include explicitly fenced, checkpoint-free preparations in the existing finite recovery candidate inventory. Normal launch, checkpoint, provider evidence and admission checks still decide whether work may resume. - Expose the existing board stop-confirmation operation for historical foreign-host owners. Bind confirmation to the exact run, boot, PID, process start and host. Reject changed ownership. - Report narrowly defined old, unowned preparations as `retainedBlockedRuns` outside the live upgrade census. Local executions, pending admission, drain state, truncation and positive checkpoint proof retain their barriers. - Update the execution contract and add PostgreSQL, real-route, race and preservation regressions. ## Verification - Genuine red/green regressions cover the old upgrade census, missing native lease registration, the volume-backed default, foreign-host confirmation, and scoped recovery skipping checkpoint-free preparation. Test bodies ran against real PostgreSQL before their corresponding production changes. - Combined verification before the final idle correction (head `055be6c84d9`): **459 passed, 22 skipped, 11 files passed**. Covers controller lease/identity, native restart classification and integration, host confirmation, heartbeat recovery, instance route authorization, upgrade readiness, and queued first-turn input. The focused command used `pnpm exec vitest run` with those 11 suites and `--maxWorkers=1 --no-file-parallelism`. The 22 skipped cases require the Rust debug fake-Codex fixture, which was not built locally; the real-process CI coverage must pass. - `pnpm --filter @paperclipai/server typecheck`: passed on head `055be6c84d9b058064daad0af71bb5d893ab9529`. - Actual route proof uses the real heartbeat service and PostgreSQL. Unauthorized access returns 403. The retained run and owner records remain unchanged. Recovery integration admits one bootstrap claim across two replacement controllers, keeps a live owner held, and preserves local execution. - `pnpm verify`: exit 2, requests broad CI because the execution contract changed. Full local suite not run. Current-head CI must supply broad coverage. - Local source checkout on macOS ARM64, Node v25.6.1, pnpm 9.15.4. Tests use isolated PostgreSQL. No live provider restart is claimed by those fixtures. - CI found an idle-query regression in the first version of the lease wiring. Two focused PostgreSQL regressions reproduced the misplaced identity-read notification and the missing claim notification. The fix moves registration to actual ownership writes. `env -u npm_execpath pnpm exec vitest run server/src/__tests__/maintenance-server-routes.test.ts --maxWorkers=1 --no-file-parallelism`: **17 passed**, including 90 real seconds and two qualified hourly windows with no extra idle queries. An initial local setup attempt inherited npm metadata and did not run tests; clearing that metadata selected the correct pnpm helper. Three focused controller/native suites also passed **168 tests, 22 fixture-gated skips**. Server typecheck passed after the correction on head `4a527326dcd5b3a9e286f0e4c3e761d3056af4d9`. - The older CI server shard also exposed a same-space graceful-restart regression. The new boot fallback had overridden local process-exit proof. Seven suspension cases and a focused unit regression reproduced it; the fix limits boot fallback to foreign or unknown PID spaces. Fresh verification on `acf1065d98150210c1e227c7f3b05a1eb489730b`: **251 passed, 22 fixture-gated skips, 4 files passed**, including all **40** graceful-restart suspension tests. Server typecheck passed. The local regression failed before this correction, and the same-space and foreign-space paths pass together after it. - Current-head CI completed successfully on `acf1065d98150210c1e227c7f3b05a1eb489730b` (run 42568109784). Greptile reviewed that exact head at 5/5 with no unresolved review threads. - Secret and private-data scans passed. No workflow, lockfile or migration changes. ## Risks - The instance lease now applies to volume-backed Railway services. Its existing 30-second renewal, quiescence, grace and self-fence behavior applies while work is held. Idle servers retain the existing no-query behavior. An explicit `PAPERCLIP_CONTROLLER_INSTANCE_LEASE=off` remains available. - A controller lease proves only the server owner's state. It does not prove an agent process stopped or authorize replay, resource release or volume removal. The existing classifier and provider/runner checks remain responsible for those decisions. - Previous-release records keep their shape and saved state. A missing boot lease stays unprovable and requires the scoped stop-confirmation path. A changed confirmation target is rejected. - The upgrade observation does not recover a held task by itself. Source tests do not qualify a hosted replacement. Deployment and the affected task's useful result must be verified separately. - This repairs recovery after a crash; it does not claim to identify or fix every source of heap exhaustion. ## Model Used OpenAI `gpt-6.1-sol`, xhigh reasoning effort, through Codex in T3 Code, with repository, shell, and GitHub tools. Context window size is not exposed by the runtime. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [x] All Paperclip CI gates are green - [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge --------- Co-authored-by: Paperclip <noreply@paperclip.ing> |
||
|
DottaandClaude Opus 5.5
|
95bf90bd2c |
fix(chat): post a conversation turn's answer while an approval holds the task (#20179)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - People talk to agents in chat. The agent's final message becomes the chat reply. > - A chat turn can end while the task stays held. For example, the agent resubmits a hire approval and the approval is pending. > - The finalizer keeps the task in review for the approval (`governed_response_waiting`). That decision is not `conversation_turn_finished`. > - Run completion then treats the yielded turn as a control-plane wait. It records `presentationDecision: none` and posts no comment. > - The person never sees the agent's answer. The chat shows nothing after the request. > - This pull request counts a committed conversation turn as answered unless its own pending question or confirmation card presents it. > - The benefit is that the person always gets the reply, whatever state the task keeps. ## Linked Issues or Issue Description Refs #19971 (post the full final reply). That fix covers the reply text. This fix covers replies that were never posted. **What happened** In a chat, the person asked the CEO agent to change a proposed hire to a cheaper model and send it back for approval. The agent resubmitted the approval correctly. It called `paperclip_finish`, and Paperclip accepted the report with "task is still waiting for approval". The run succeeded and the task moved to `in_review`. The run event `run.presentation.resolved` recorded `commentAction: none` (reason `yielded_control_plane_wait` on master). The agent's final message never reached the chat. **Expected behavior** A chat turn's accepted final answer is always posted to the conversation. An approval, an execution stage or a dependency blocker can hold the task. It must not hide the reply. Only a question or confirmation card that the turn itself left pending presents the turn instead of its prose. **Steps to reproduce** 1. Open an Agent Chat with an agent. Link a hire approval in `revision_requested` to the chat task. 2. Ask the agent to change the hire and resubmit it. 3. The agent resubmits the approval and ends the turn with a yielded result and a response wake. 4. See that the chat shows no reply. The run is `succeeded` and its presentation decision is `none`. ## What Changed - `server/src/services/native-runtime/native-governance.ts`: move the pending-card query of `pendingNativeGovernance()` into `pendingNativeInteraction()`, with no change in behavior. Add `conversationTurnAnswered()`. It returns false only when a pending thread interaction (the same predicate as status arbitration) gates the turn. It checks cards on their own, so a pending execution stage does not hide a card. - `server/src/services/heartbeat/run-completion.ts`: a successful conversation run whose status decision committed now passes `conversationTurnFinished: true` to the presentation resolver, unless `conversationTurnAnswered()` is false. The existing `conversation_turn_finished` path is unchanged. - `server/src/services/heartbeat/run-completion.test.ts`: add a regression table with a real database. A pending approval, a revision-requested approval, a pending execution stage and a dependency blocker now post the answer. A finished turn still posts. The turn's own question card still owns the response, also when an execution stage is pending. - `doc/execution-semantics.md`: state the rule. Other finish outcomes I checked: - `done` or `needs_review` with a gate (`governed_gate_pending`): the result is not yielded, so the reply was already posted. - `blocked` (`paperclip_block`, waiting on a person or another task): not yielded, so the reply was already posted. - Yielded with a dependency blocker (`durable_dependency_blocker_bound`), or with `record_finalization_error` (`prior_status_preserved_no_live_path`, `monitor_wait_authority_lost`): the reply was also dropped. This fix covers them, because the rule does not depend on the reason code. ## Verification - `cd server && npx vitest run src/services/heartbeat/run-completion.test.ts`: 118 passed. Before the fix, 4 new cases failed (both approval states, execution stage, dependency blocker). The finished-turn and question-card cases passed both before and after. - `cd server && npx vitest run src/__tests__/heartbeat-run-summary.test.ts src/services/heartbeat/run-comment-selection.test.ts src/__tests__/slack-conversation-lifecycle.test.ts src/services/native-runtime/agent-chat-response-recovery.test.ts src/__tests__/chat-run-milestone-governed-work.test.ts src/__tests__/agent-conversations.test.ts`: all passed. - `tsc --noEmit -p server`: no errors in the changed files. Unrelated errors in this worktree come from an unbuilt local Runner package. - An end-to-end chat eval run on the case branch is in progress. I will add its result here. ## Risks - Low risk. The change only affects successful Agent Chat runs with a committed status decision. Ordinary tasks and external chat (Slack, GitHub) keep their own paths. External chat returns before this check. - A turn that yields while an approval is pending now posts its prose. The approval stays pending and the task status does not change. - One extra small read (the task row and pending gates) runs after a conversation turn that is not `conversation_turn_finished`. It runs after the provider turn ends, so it does not add time before the first token. - Previous-release state: no persisted shape changes. The fix reads `finalizationPhase` and `finalizationReasonCode`, which the previous release already writes. A run in flight at the deploy completes on the new code and posts its reply. A run that completed before the deploy keeps its recorded decision; nothing re-presents it. - Known gap, unchanged: background presentation recovery for conversations still requires `conversation_turn_finished` with a `done` result. If the server restarts between the status commit and the live presentation, a held yielded turn is not re-posted. That is a separate follow-up. > For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and discuss it in `#dev` before opening the PR. Feature PRs that overlap with planned core work may need to be redirected — check the roadmap first. See `CONTRIBUTING.md`. ## Model Used - Anthropic Claude Opus 5.5 (`claude-opus-5-5`), with tool use (shell, file edits, test runs) in Claude Code. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
5a51a7685d |
fix(budgets): raise_budget_and_resume continues the work its budget stopped, once (#20144)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - A budget hard stop cancels the scope's running work; the board resolves it with Raise budget and resume > - In the restart-chaos H1 scenario, the cancelled tasks did not continue after resume and each needed a manual Retry > - Resume is the board's explicit decision to continue, so the work it stopped should continue once, as if the board clicked Retry > - This pull request files that Retry automatically, keyed by the cancelled run, once the run's stop has settled > - The benefit is that a budget raise restores the work without manual Retries, without repeating completed work, and without charging the run's attempt budget ## Linked Issues or Issue Description Refs #20143 (bug B; this PR builds on it), #20113 (bug A), #20101 (superseded: split into #20143 and this PR) **What happened?** Restart-chaos waves w-20261108T195944-s42 and w-20261108T210118-s9802 (H1). After `raise_budget_and_resume`, CHA-3's budget-cancelled run (enforcement stop) and CHA-1's run (agent lifecycle pause from the budget hold) stayed cancelled until someone clicked Retry. **Expected behavior** Resuming continues each task its budget stopped, exactly once, with no attempt charged, an activity entry, and no repeat of work that already continued. A run a person stopped stays stopped. **Steps to reproduce** Set a budget below spend while a task runs; let the stop cancel the run; resolve with Raise budget and resume. Before this change, no new run is queued. ## What Changed - New table `budget_resume_claims` (migration 0451). One row per cancelled run, outside the run's result JSON, so a Stop that settles during the resume cannot erase it. - `markBudgetCancelledWorkForResume` (called by `resolveIncident` before the incident closes): claims every task run its budget stopped since the incident opened. That covers runs budget enforcement fenced, including stops still in flight, and runs the agent lifecycle pause stopped for this budget (`pauseStop`, carried from the lifecycle operation's `pauseCause` so a resume before the Stop keeps the link). Runs a person stopped are excluded. Two indexed passes led by `(company_id, status)`. - `deliverBudgetResumeContinuations`: once a claimed run's stop settles, files the resolving board member's Retry (`requestTaskRetry`, idempotency key `budget-resume:<run>`, with the member's session source and the chat Retry callback). The Retry path checks, under the task lock, that the run is still the task's latest and the task is open, so continued or completed work is never repeated. Claim settlement and its activity entry commit together; retryable refusals and errors retry. - Delivery runs in the always-on budget-enforcement worker after each pass's Stop delivery. Its schedule includes deliverable claims and the earliest stale-claim expiry. The execution-continuation worker is a backstop. - `agent.paused` / `agent.resumed` logging is in #20113. `budget.cancelled_work_resume_requested` (incident) and `budget.cancelled_work_continued` / `_not_continued` (task) are recorded here. ## Verification - New `budget-resume-continuation.test.ts` (Postgres, real `resolveIncident` → claims → `requestTaskRetry` → real `resumeTaskRetry` into a queued run): - each task continues once, including a blocked task; repeat delivery or resolve saves nothing new; - a stop still in flight at resume, whose terminal write replaces the result JSON, is still continued; - a 30-task scope is fully continued; - a lifecycle-pause stop caused by the budget is continued, and one caused by a manual pause is not; - continued work, done tasks, a run that finished before its stop, and person-stopped runs (including ones enforcement also fenced) are not continued. - `lifecycle.test.ts`: the pause records its cause on the operation and on stopped runs, including when the budget is resumed before the pause's Stop. `budget-enforcement-worker.test.ts`: delivery after Stop delivery, and a wake at claim expiry. - 287 tests across the affected suites passed. Server typecheck (`typecheck-layers`, low-memory): no errors. - Not run: the H1 chaos rerun (Phase 3 runs it after merge). ## Risks - Resume now continues budget-stopped tasks automatically. A board that wants a task stopped can Stop it again; person-stopped runs are never continued. - A new table and one index (migration 0451). > For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and discuss it in `#dev` before opening the PR. Feature PRs that overlap with planned core work may need to be redirected — check the roadmap first. See `CONTRIBUTING.md`. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, via Claude Code with tool use (shell, file edits, test execution). Extended reasoning enabled. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
b54317d3fb |
fix(native): give a preparing run's first turn the user's queued messages instead of interrupting it (#20168)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - A Paperclip Runner (native) run prepares its workspace and
environment before it starts its first provider turn. This can take
about a minute.
> - A user's message that arrives during that time waits as a deferred
wake behind the run. The Interrupt control then cancelled the run before
any provider turn.
> - The cancel threw away the whole preparation. A new run had to
prepare again, so the user waited longer for the first token.
> - This pull request gives the run's first turn the queued messages.
The run seals its first-turn input immediately before `turn.start`, so
the messages are part of the prompt that the turn starts with.
> - The benefit is that the user sees the message handled in the same
run, with no cancel and no second preparation.
## Linked Issues or Issue Description
**What happened?**
A Paperclip Runner run took about 54 s to prepare (environment startup,
workspace, runtime selection). The user sent a second message 32 s into
that time and clicked Interrupt 21 s later. Interrupt cancelled the run
as `operator_interrupted` ("Interrupted to send queued messages") before
any provider turn. A new run then prepared again from the start. The
runtime-selection transaction that froze the run's input completed about
5 s before the click, and the coordinator claim came 4 s after it, so
the message could not be given to the first turn at any point in the old
code.
**Expected behavior**
When a user message arrives for a task whose current run has not started
its first provider turn, the run does not stop. The message goes into
that run's first turn. A run that is in a turn is interrupted as before.
**Steps to reproduce**
1. Assign a task to a Paperclip Runner agent with a sandbox environment,
so preparation takes a long time.
2. While the run prepares, post a second comment on the task.
3. Click Interrupt on the queued message before the run starts its
provider turn.
4. See that the run is cancelled and a new run prepares again.
**Paperclip version or commit**
Staging revision
|
||
|
DottaandClaude Opus 5.5
|
098f7d370b |
fix(workspaces): release an unchanged remote writer without a snapshot and wake its waiters (#20171)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - A remote (sandbox) task workspace has one writer run at a time. The
writer holds the workspace until its files are saved as a snapshot
> - A run that a user comment interrupts right after its claim did no
work. No provider turn started and no command ran in the sandbox
> - Paperclip still kept its sandbox, then took a full snapshot of the
unchanged workspace. On a 2 GB repository this took 3 minutes
> - The next runs failed setup with `workspace_busy` ("ended, but its
workspace save has not settled"). The retry then waited for the
one-minute settling re-check after the save. The user's message waited
more than 4 minutes
> - This pull request releases such a writer at once, at the saved
version it started from, when durable evidence proves that it changed
nothing. It also wakes parked `workspace_busy` waiters when a hold
settles
> - The benefit is that an interrupted run no longer delays the next
message by a full workspace save
## Linked Issues or Issue Description
**What happened?**
A remote-workspace run was claimed (`native.coordinator.claim`). One
second later a board comment interrupted it (`operator_interrupted`). It
made no provider turn and no tool calls. Cancelled-workspace
preservation still retained its sandbox, and workspace recovery took a
full `remote_workspace_snapshot` of the unchanged workspace (3 minutes
for a 2 GB repository). Two follow-up runs were cancelled with
`workspace_busy`. Their retry was due during the save, and it parked for
the one-minute settling re-check after the save finished.
**Expected behavior**
A writer that provably changed nothing releases the workspace at once
and keeps its existing `snapshotId`. The next run is admitted without
`workspace_busy`. When a save does run, waiters start as soon as it
commits.
**Steps to reproduce**
1. Run a task on a remote sandbox workspace that already has a saved
version.
2. Interrupt the run with a board comment right after the claim, before
the provider turn.
3. See the full snapshot operation on the interrupted run and
`workspace_busy` on the follow-up runs.
**Paperclip version or commit**
Staging served revision
|
||
|
Devin FoleyandPaperclip
|
546cfee919 |
fix(github): send gh repo create and fork to the connection's tools (#20142)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Agents with a managed GitHub connection get `git` and `gh` launchers that use the connection's credential > - The launcher gives that credential to any `gh` command, and the broker does not see which command runs > - So `gh repo create` and `gh repo fork` skip the connection's tool switches, ask-first approvals, owner defaults, and audit > - The connection's `create_repository` and `fork_repository` tools apply all of these > - This pull request stops `gh repo create`, `gh repo new`, and `gh repo fork` when a managed credential is in use, and names the tool to use instead > - The benefit is that repository creation follows the connection's rules, whichever path the agent tries first ## Linked Issues or Issue Description Refs #20141 (default owner for `create_repository` and `fork_repository`). **What happened?** An agent with a managed GitHub connection can run `gh repo create` in its shell. The command uses the connection's token, but it does not use the connection's tool rules. If the user turns off or reviews `create_repository` on the connection, the agent can still create a repository with `gh`. The owner default from #20141 also does not apply, so the command can fail with `403 Resource not accessible by integration` when the App is installed only on an organization. **Expected behavior** With a managed connection, repository creation and forks go through the connection's tools. The tool switch, approvals, owner default, and audit apply. **Steps to reproduce** 1. Connect GitHub with a managed connection. 2. Set the connection's `create_repository` tool to off or ask first. 3. In an agent run, run `gh repo create demo --private`. 4. The command runs with the connection's credential and ignores the tool setting. ## What Changed - `packages/adapter-utils/src/github-launcher.ts`: when the broker returns a managed credential, `gh repo create`, `gh repo new`, and `gh repo fork` exit with code 1 before `gh` starts. The message names the `create_repository` or `fork_repository` tool and tells the agent to finish with Git. - `--help` still runs. Other `gh` commands do not change. Executions without a managed credential run these commands as before. - `doc/execution-github-identity.md`: describes the behavior and states that this is guidance, not a security boundary. ## Verification - New test in `packages/adapter-utils/src/github-launcher.test.ts`: "sends repository creation and forks to the connection's tools when a connection is in use". It covers `create`, the `new` alias, `fork`, `--help`, another `gh repo` command, and the case without a connection. It fails without the launcher change. - `npx vitest run src/execution-target.test.ts src/github-launcher-environment.test.ts src/github-launcher.test.ts` in `packages/adapter-utils`: 83 passed. - `npx vitest run src/drivers/acpx/github-launcher.test.ts` in `packages/paperclip-runner`: 3 passed. - `npx vitest run src/services/heartbeat-github-launchers.test.ts` in `server`: 12 passed. - `npx tsc --noEmit -p packages/adapter-utils` passes. ## Risks - Agents that used `gh repo create --source . --push` must now call `create_repository`, then add the remote and push with Git. The message tells them to do this. - If the connection's tool is off for the agent, the agent cannot create a repository. This is the purpose of the tool setting. - This is not a security boundary. Other commands that receive the token, such as `gh api`, can still call the same GitHub endpoints. - Running launchers keep their earlier source until the next run stages new launchers. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, used through Claude Code with tool use (code reading, shell commands, and test runs). ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge Co-authored-by: Paperclip <noreply@paperclip.ing> |
||
|
Devin FoleyandPaperclip
|
0e49576fd9 |
fix(github): default repository owner to the only installed organization (#20141)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Agents use a managed GitHub connection to create repositories and to fork repositories > - The Paperclip GitHub App can be installed only on an organization, and not on the user's personal account > - Then `create_repository` and `fork_repository` fail when `organization` is empty, because GitHub sends them to the personal account > - Agents, approval cards, and connection tests all fail on their first attempt with `403 Resource not accessible by integration` > - This pull request fills an empty `organization` with the only organization where the App is installed > - The benefit is that repository creation works on the first attempt for these connections ## Linked Issues or Issue Description No public issue exists. Refs #19655 (GitHub repair guidance for the same error). **What happened?** A user connected GitHub through the managed connection. The Paperclip GitHub App was installed on one organization and not on the user's personal account. An agent called `create_repository` without `organization`. GitHub returned `POST https://api.github.com/user/repos: 403 Resource not accessible by integration`. The test button on the connection page gave the same error. The agent then stopped and asked the user to create the repository manually. **Expected behavior** When the App can create repositories only in one organization, an empty `organization` uses that organization. The approval card and the test result show that owner. **Steps to reproduce** 1. Install the Paperclip GitHub App on one organization only. Do not install it on the personal account. 2. Connect GitHub with the managed connection. 3. Run `create_repository` with only `name`, from an agent or from the connection test dialog. 4. GitHub returns 403. ## What Changed - Added `server/src/services/tool-access/github-default-owner.ts`. It finds the default organization from the grant's access snapshot. It returns an organization only when there is exactly one installation, and the snapshot marks that owner's repositories as organization repositories. Older snapshots without account types get no default until access is refreshed. - `governedToolArguments` in `tool-gateway.ts` fills an empty `organization` for `create_repository` and `fork_repository` on managed GitHub connections. The change occurs before policy, approval signing, previews, and audit. Thus agent calls, approval cards, and connection tests use the same owner. - The grant lookup runs only for these two tools, and only when `organization` is empty. If the credential is unavailable, the call does not change and fails with its own error at dispatch. - An explicit owner does not change. A fork into its own owner does not change. Token (PAT) connections, personal installations, and multiple organizations keep GitHub's behavior. - `projectedConnectionToolInputSchema` describes the default on the `organization` argument. Agents and the connection test form see where the repository goes. - Updated `doc/connections/GITHUB.md`. ## Verification - `pnpm --filter @paperclipai/server typecheck` passes. - `npx vitest run src/services/tool-access/github-default-owner.test.ts`: new unit tests for the owner rules (single organization, personal installation, several organizations, another person's account, older snapshots, blank values, forks into the same owner, schema text). - New gateway test in `src/__tests__/tool-gateway.test.ts`: "defaults GitHub repository creation to the only installed organization for agents and connection tests". It covers an agent call, an explicit owner, a connection test call, and a personal installation. It fails without the gateway change. - Related suites pass, 393 tests in 9 files: `tool-gateway.test.ts`, `tool-gateway-service.test.ts`, `connection-instructions.test.ts`, `github-access-errors.test.ts`, `github-default-owner.test.ts`, `tool-gateway-discovery-http.test.ts`, `github-operation-credentials.test.ts`, `project-git-authorization.test.ts`, `prompt-injection-gateway.test.ts`. - Not tested against a live GitHub App installation. ## Risks - Low risk. The default applies only when the grant snapshot shows one organization installation and no personal installation. Without the change, these calls fail with 403. - A stale access snapshot can choose the organization after a user installs the App on the personal account. Installation events and **Refresh access** update the snapshot. - An approval that was created before this change, without `organization`, no longer matches the projected arguments. That approval cannot run and must be requested again. Without the change, it fails with 403. - `gh repo create` and `gh repo fork` in agent shells do not use the gateway, so they keep GitHub's personal default. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, used through Claude Code with tool use (code reading, shell commands, and test runs). ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge --------- Co-authored-by: Paperclip <noreply@paperclip.ing> |
||
|
DottaandClaude Opus 5.5
|
95d476c57b |
docs(plans): record the measured seed experiment for per-turn uploads (#20139)
## Thinking Path - #20100 proposed sending only the changed files on per-turn follow-ups, with the repository seed held inside Daytona in a snapshot or a volume. It asked for one measured experiment before building. - Snapshot and volume operations were blocked by our Daytona key's permissions. A restricted key with those permissions now exists, so the experiment has run. - This PR records the numbers and the decision in the design doc. ## Linked Issues or Issue Description Refs #20100 **What happened?** The #20100 design's seed-location choice was unmeasured: snapshot build time and volume read speed were both open questions. **Steps to reproduce** 1. Run the SDK timing of direct upload, volume extract, and snapshot build and create on the paperclip seed (58.7 MB gzip) in Daytona `us`, 2 vCPU. **Expected behavior** The design doc states the measured winner, the per-turn saving, the one-time cost, and the work needed to build it. ## What Changed - `doc/plans/2026-11-08-incremental-per-turn-workspace-upload.md`: - The status now reads: measured; snapshot chosen; not built. - New "Experiment results" table: - direct upload: 6.4 s per turn; - volume: 6.8 s per turn, which rules it out; - snapshot: about 2–3.5 s per turn, after a 177 s build per seed generation. - Findings: fork is unsupported, compression is exhausted, and the snapshot builder runs as `daytona` and does not extract archives. - New "Work to build the snapshot seed" list: lifecycle, sandbox creation, verification and fallback, expiry and cleanup, and rebuild policy. ## Verification Docs only. Every Daytona resource the experiment created was deleted, and an account-wide listing confirmed none remained. ## Risks None at runtime. ## Model Used Claude Opus 5.5 (`claude-opus-5-5`, Anthropic), 1M context window, tool use and code execution, via Claude Code. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
2a16be36ab |
docs(plans): per-turn follow-ups without re-uploading the repository (#20100)
## Thinking Path - Per-turn remote sandboxes stay the default while warm reuse (#20048) is on hold. - On the paperclip clone, a per-turn follow-up spends about 13 s in `workspace_sync.transfer`, though its changes pack in 0.5 s: every fresh sandbox receives the whole 70 MB (gzip) seed again, at about 5 MB/s from the controller. - The coordinator asked for a design that uploads only changed files into a fresh sandbox, with a full-upload fallback, measured against the spans. It also asked to build it only if the implementation is bounded and safe. ## Linked Issues or Issue Description Refs #20048 **What happened?** Per-turn follow-ups on large repositories wait about 13 s re-uploading the full workspace seed even when almost nothing changed (pack 0.54 s, transfer 13.3 s, time to first token about 31 s). **Steps to reproduce** 1. Run `tests/restart-chaos/ttft-baseline.ts --env daytona --workspace-repo <paperclip clone>` with per-turn sandboxes. 2. Compare `workspace_sync.pack` and `workspace_sync.transfer` on the follow-up turn. **Expected behavior** A follow-up uploads only its overlay into a sandbox that already holds the seed, falling back to the full upload when unsure. ## What Changed - `doc/plans/2026-11-08-incremental-per-turn-workspace-upload.md`. It contains: - the measured span table and the size analysis (220 MB archive, 70 MB gzip, about 5 MB/s); - the expected saving: about 12–13 s per follow-up, time to first token about 31 s → about 18 s; - a design keyed by seed generation with a content-hashed manifest, using a Daytona snapshot (preferred) or a Daytona volume to hold the seed; - the overlay-only turn flow, with a verified fallback; - one bounded experiment to choose the seed location before building. - Not built yet: both seed locations need provider capabilities (snapshot build time, volume read speed) that must be measured first, so the implementation isn't bounded until the experiment picks one. ## Verification Docs only. ## Risks None at runtime. ## Model Used Claude Opus 5.5 (`claude-opus-5-5`, Anthropic), 1M context window, tool use and code execution, via Claude Code. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
c02fd8d427 |
docs(plans): salvage-then-retire design for kept restore-source and done-task warm sandboxes (#20087)
## Thinking Path - Paperclip keeps some remote sandboxes after their run: warm ones for reuse, and failed runs' stopped sandboxes as workspace-restore sources. - In the warm Daytona chaos waves for #20048, the real leaks were restore sources that nothing retires once a user Retry closes their recovery actions. Separately, a done task's suspended warm sandbox is only destroyed when its environment is deleted. - Retiring them early needs a provider-I/O salvage step: compare against the durable save, export anything unsaved to the issue, then destroy. That spans recovery, storage, attachments and teardown. The coordinator asked for a design note so it can be its own reviewed task instead of a rushed patch. ## Linked Issues or Issue Description Refs #20048, #20085, #20086 **What happened?** Closed restore sources and done-task warm sandboxes stay suspended until their environment is deleted, costing storage. No path compares them with the last durable save or exports their unsaved files before retirement. **Steps to reproduce** 1. A run fails; its sandbox is kept as a `retain_on_failure` restore source. 2. The user clicks Retry; a fresh run succeeds and the task is done. 3. The restore-source sandbox is never retired until the environment is deleted. **Expected behavior** After a clean compare, or a confirmed export of anything unsaved to the issue, the sandbox is destroyed. If salvage fails, it stays suspended and is retried. ## What Changed - `doc/plans/2026-11-08-restore-source-salvage-and-warm-retirement.md`: a design for one owned kind (`environment_kept_sandbox_retirement`) on the shared reconciler. It covers listing rules for both cases (24 h default retention for done-task warm sandboxes), a three-stage salvage state machine (compare, export, retire) with a provider-I/O step and receipt table, failure handling that never destroys without proof, and the test plan. ## Verification Docs only; no code or tests changed. ## Risks None at runtime. Additive documentation. ## Model Used Claude Opus 5.5 (`claude-opus-5-5`, Anthropic), 1M context window, tool use and code execution, via Claude Code. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
b32ef9f111 |
perf(budgets): judge time to first visible agent activity; keep time to first text as a reported metric (#20077)
## Thinking Path - The native speed budgets (plan invariant 8) judge how fast an agent responds. They judged `task.time_to_first_token`: time to the first deliverable **text**. - In the clean local restart storm (w-20261108T153612-s9401), Claude breached its TTFT budget (cold p50/p95 17.4/34.7 s against 16.5/30). The per-run spans showed this was not a control-plane wait: in all 49 cold runs (and 76% of warm), Claude started a tool call before writing any text. Its first visible activity came at 7.4 s p50 from run creation; first text came at 19.6 s. Codex never calls a tool first. - So first text charges model tool work as latency for tool-first adapters. What the user experiences as "the agent responded" is the first visible activity: a text delta, a tool-call start, or a thinking indicator. - The coordinator decided: budgets judge time to first visible agent activity from run creation, with the budget numbers unchanged; first text stays a reported metric with no budget for tool-first adapters. ## Linked Issues or Issue Description **What happened?** Local storm s9401: the Claude TTFT budget breached on every aggregate, while the control-plane steps (queue 0.05 s, prepare 0.3 s, runner startup ~4 s, first provider event 1.3–1.8 s) were well within budget. The rest of the first-text wait was Claude streaming tool input (about 5.5 s per call) and running tools before it wrote text. **Steps to reproduce** 1. Run a Claude task whose first action is a tool call (most real tasks). 2. Read the run log: `provider.time_to_first_agent_event` ends at the tool call; `task.time_to_first_token` ends many seconds later at the first text. 3. The speed budget judges the later one and reports a breach. **Expected behavior** The budget judges the first visible agent activity from run creation (`task.time_to_first_agent_activity`). Time to first text is still recorded and reported. ## What Changed - `server/src/services/native-runtime/native-session-executor.ts`: when the first meaningful agent event commits (the existing `firstMeaningfulAgentEventKind`: text delta, tool-call start, or reasoning), record `task.time_to_first_agent_activity`. It uses the same detached anchor resolution as TTFT (accepted comment, answered question, durable wake request, or run creation), so it adds nothing to the turn path. - `server/src/services/native-runtime/native-speed-budgets.ts`: the metric is now named `time_to_first_activity` (was `ttft`), with identical budget numbers. The type comment documents the definition and why first text is unbudgeted. - `server/src/services/native-runtime/native-run-trace.ts`: the budget check judges `task.time_to_first_agent_activity` from run creation, keeping the existing input-wait handling. `task.time_to_first_token` is no longer budgeted. - `tests/restart-chaos/ttft-baseline-stats.ts` and `ttft-baseline.ts`: report first activity from run creation (the budgeted value) next to TTFT. - Docs: `doc/observability.md` (new span) and `doc/plans/2026-11-07-ttft-baseline.md` (metric table). - Tests: - A tool-first turn records first activity at the tool call (5 s) while TTFT records first text (16 s). - A tool-first trace with slow first text but fast first activity is within budget. - The existing budget tests now judge the activity span. Observability path (AGENTS.md section 5.7): the new run-log span uses only the existing `eventKind` and `anchor` attributes. Its name survives run-log redaction intact; checked with `redactEventPayload`. The operator OTLP gate is unchanged. It is not Telemetry. ## Verification - `cd server && npx vitest run src/services/native-runtime/native-speed-budgets.test.ts src/services/native-runtime/native-run-trace.test.ts`: 44/44 pass. - `cd server && npx vitest run src/services/native-runtime/native-session-executor.test.ts -t "first visible activity|first assistant text once|first-token timing"`: 7/7 pass. - `npx vitest run --root tests/restart-chaos`: 159/159 pass; `tsc` on `tests/restart-chaos` is clean. ## Risks - **A metric rename.** Consumers of `run.performance.budget_exceeded` see `metric: "time_to_first_activity"` instead of `"ttft"`. In-repo there are none besides the tests; Phase 3's chaos speed grade switches to the new span. - **A looser budget for slow first text.** Intentional: first text is reported, not judged. A regression in control-plane latency still shows up in first activity. ## Model Used Claude Opus 5.5 (`claude-opus-5-5`, Anthropic), 1M context window, tool use and code execution, via Claude Code. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
4f7eee10c2 |
feat(recovery): let the board confirm an unrecorded owner stopped for every settling process kind (#20084)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - Recovery releases in-progress records (leases, claims, intents) only
when it can prove their owner is gone
> - A record saved before its server process was fully recorded is
`unprovable`, and it is never released on its own
> - #20067 added a board-only action to confirm that such an owner
stopped, but only the environment run lease used it
> - Other settling kinds (for example the legacy controller lease, which
shows `missing_identity` rows in the Phase 2 reconciler data) stay stuck
with no way forward
> - This pull request opens the same confirmation to every settling kind
whose record names a partly recorded server process
> - The benefit is that an operator can release these stuck records
safely, and recovery settles them at once
## Linked Issues or Issue Description
Refs #20067, #20054
**What happened?**
The Phase 2 reconciler data shows `legacy_controller_lease` records that
stay `unprovable` with `missing_identity`: they were saved before
Paperclip stored the controller's PID space. Recovery never releases an
unprovable owner, and only the environment run lease offered the board's
"confirm the previous server stopped" action. These records, and records
of other settling kinds saved the same way, had no path forward.
Also, a board confirmation did not wake the reconciler. Unprovable
records are not watched, so the release waited for some unrelated pass.
**Expected behavior**
Recovery asks the board on the record's task. After the board confirms,
recovery releases the record through the kind's own compare-and-swap
settle, at once. A confirmation never releases a record that changed
since the board was asked.
**Steps to reproduce**
1. Have a legacy run whose `controller_process` was recorded without a
host hash (an older server).
2. Run the ownership reconciliation pass. The record is `unprovable`
(`missing_identity`) and no board action appears.
**Paperclip version or commit**
master at
|
||
|
DottaandClaude Opus 5.5
|
e8a3f1f6f2 |
feat(recovery): prove a replaced container's records gone through a fenced controller epoch (#20054)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - After a restart, the control plane settles in-progress records (leases, claims, intents) only once it proves their owner process is gone > - The proof compares host and PID space, so a stamp from a replaced container or a rebooted machine stays unprovable forever, and its records never settle > - Self-hosted users on Docker, Kubernetes and Fly hit this on every redeploy; a Railway deployment with an exclusive volume is covered by `deployment_replaced` (#20026) > - This pull request adds a fenced controller epoch: each boot holds a database lease, and an expired boot is fenced so its records can be proven gone > - The benefit is that work stranded by a replaced container is recovered, without keeping an idle database awake ## Linked Issues or Issue Description Follow-up to #20026 and the Phase 2 durable-ownership data. Related: #19965 (owner stamps), #19883 (process space). No duplicate PR was found. **What happened?** A record stamped by a server process in another container reports `different_process_space`. A dead PID elsewhere is never proof, so the record stays `unprovable` and is never settled. Every redeploy of a self-hosted container leaves such records behind. **Expected behavior** A record owned by a server boot that was replaced (stopped cleanly, or lost its lease and was fenced) is proven gone and settled exactly once. A live boot's records stay with it. An idle server issues no recurring queries. **Steps to reproduce** 1. Run Paperclip in a container and start work, so an environment lease is stamped by that boot. 2. Replace the container. The new container has a new PID namespace. 3. The ownership reconciler reports `different_process_space` (unprovable) for the old lease, forever. **Paperclip version or commit** `master` before this PR. ## What Changed - New `controller_instances` table (migration 0449): boot id, epoch (bigserial), process space, pid, a lease on the database clock, `fenced_at` / `fenced_by_epoch`, `stopped_at`. - `controller-instance-lease.ts`, registered with the shared delivery-work coordinator (`controller-instance-lease` queue, runs during drain): - **Holds work → renews; idle → silent.** One upsert creates the row on first use and extends the lease only if the boot holds work: a stamp within the last 90 s, or a stamp whose ownership write may still be pending (held until the server's tracked work settles, capped at 10 minutes; a renewal that sees some other held record does not clear it, so a slow ownership write cannot be outlived), or an in-progress record held by its boot id. Those records are active environment leases, in-flight cleanup claims, sandbox creation intents, workspace finalize attempts, agent lifecycle steps, native finalizer and cleanup leases (`lease_holder` bound to the token), open native coordinator claims, and running legacy runs. Renewal runs every 30 s while the boot holds work. When it finds none, the lease lapses and the worker sleeps until the next stamp (`onControllerStamp`). - **No idle trailing statements.** When the server's last tracked work settles, the worker re-checks holdings at once, at most every 5 s (the rate limit also absorbs the worker's own run). So the last check happens with the work, not 30 s later. - **Quiesce, don't crash.** Each renewal has a 10 s bound. A failed or timed-out renewal applies an internal task drain, so no new work is dispatched. A recovered renewal resumes. The boot exits only after 60 s of failure (counted from the first failed attempt's start) while it still holds work, or at once when it finds itself fenced. - **Fence on demand, by a newer epoch.** For a stamp from another host or PID space, the prover reads that boot's row: - `stopped` or `fenced`: gone. - live lease: alive. - otherwise, the boot's last sign of life is the later of its lease expiry and the record's last write plus the TTL. Within 30 s of that, the owner stays `alive` (`controller_lease_grace`), so the settle watch keeps rechecking. After that, the prover fences the boot, but only with a lease row of its own (created on demand) and only an older epoch. - A graceful shutdown records `stopped_at`. - A reconciliation pass that settles nothing (`record: false` or `settle: false`) never fences and never creates the boot's row. - `owner-proof.ts`: these proofs apply only to server-held records. A stamp with a recorded process group stays unprovable, because agent processes can outlive their server. The reconciler passes each record's `updatedAt` to the proof. - `task-admission.ts`: - the quiesce drain never replaces an operator's drain, and has no idle owner, so HTTP stays open; - a `subscribeIdleWorkSettled` hook fires when the last tracked work finishes. - Off on a Railway deployment with an exclusive volume (it has its own replacement proof). `PAPERCLIP_CONTROLLER_INSTANCE_LEASE=on|off` overrides this. A stateless Railway service keeps the lease. A server without its own lease row never fences. - `maintenance-server-routes.test.ts`: in the post-run quiet window, the lease may check once more that the run's records were released. The test now excludes those statements, then asserts that the lease issues nothing in a further window. - Docs: `doc/execution-semantics.md`; the Phase 2 data doc records the two known gaps. ## Verification - `controller-instance-lease.test.ts` (15 cases): - a replaced container with all Railway variables empty is fenced and settled exactly once over two passes; - a live lease is alive; - an owner in grace, and work written after an idle lapse, stay alive and watched; - an older epoch never fences a newer one; - no fencing without its own lease row (off or unregistered), and an idle prover creates its row on demand; - a stopped boot is gone; - process-group stamps stay unprovable; - idle issues zero statements, and one upsert renews only while records are held; - **a 45 s outage:** quiesced, no exit, resumes; - **a renewal that never answers:** bounded, and the boot exits on time; - **a 70 s outage:** exits at 60–65 s, and the successor recovers the work exactly once; - fenced: stops; - the settle re-check; - graceful stop; off on a volume-exclusive Railway deployment, on for a stateless one, and the on/off override; - a running legacy run keeps the lease live. - Red, with the proof's instance check removed: the replaced-container, live, stopped and successor cases fail. - Wider: task-drain, idle-sleep, idle-local-work, owner-proof and ownership suites, plus the maintenance quiet-window tests. Server low-memory typecheck is clean. ## Risks - **Clock skew.** The proof compares a record's `updatedAt` (often the writer's clock) with the lease (database clock). The 30 s grace absorbs ordinary skew. - **A new exit path exists.** A server whose database is unreachable for more than 60 s while it holds work now exits, after first quiescing. This is the intended safety backstop. - **Bounded trailing activity.** A lease that holds work issues one statement every 30 s, plus at most one check when its work settles. An idle server issues none. - **Boots from before this release** have no row, and their records stay unprovable. The board "mark owner gone" action for those follows in a separate PR. ## Model Used - Anthropic Claude Opus 5.5 (`claude-opus-5-5`), via Claude Code with tool use. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Nicky Leach
|
17ad23aafe | fix(workspaces): back off failed terminal cleanup and retire abandoned reservations through remove (#20020) | ||
|
DottaandClaude Opus 5.5
|
c966334ec3 |
fix(recovery): settle a run-less lease once its stamped holder is proven gone (#20026)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - The control plane recovers in-progress records after a server restart through a shared ownership path > - Run-less sandbox leases (login, probe, harness verification) were released only by a five-minute idle sweep > - That sweep releases by age: it leaves a dead holder's sandbox for minutes and takes a live login session's sandbox > - This pull request settles a run-less lease only when its stamped holder is proven gone > - The benefit is no stuck sandbox after a restart, and no live session loses its sandbox ## Linked Issues or Issue Description Related: #19965 (named the owner of login/probe leases), #19964 (the run-lease settle this mirrors). No duplicate PR was found. **What happened?** The final Daytona chaos wave `w-20261108T083036-s9401` recorded I2 `environment_login_probe_lease` (`pid_dead`) on 2 records. These were the codex agent's harness-verification leases. A restart killed their holder. The leases stayed `active` until the five-minute idle sweep released them. The same sweep releases any run-less lease after five idle minutes, also when its holder is alive. **Expected behavior** A run-less lease is released as soon as its stamped holder is proven gone. A lease whose holder is alive is never released, however long it is idle. **Steps to reproduce** 1. Start a Daytona wave: `node cli/node_modules/tsx/dist/cli.mjs tests/restart-chaos/main.ts --env daytona --daytona-image <image>` (K1/K4 faults). 2. Let a fault kill the server while an agent's harness verification holds its sandbox lease. 3. See the reconciler report I2 `environment_login_probe_lease` until the idle sweep releases it. **Paperclip version or commit** `master` at `8475826d2d`. ## What Changed - New `settleEnvironmentLoginProbeLease`, registered on `environment_login_probe_lease`. It runs only for a stamped holder that is proven gone, and compares and swaps on that exact stamp. - It moves the lease to `pending_cleanup` with the recovered-orphan reason, so the cleanup sweep destroys the sandbox in its next tick. - It leaves the lease alone when a legacy shutdown fence applies, or when another live lease holds the same provider resource. - The idle sweep for run-less leases now covers only legacy rows saved before holders were stamped (no `leaseController` key). Stamped rows are never released by age. - `holderUnchanged` is now exported from the run-lease settle so both settles share it. - The controller stamp (`leaseController`, and every other `ControllerProcess` stamp) now records the platform deployment when the platform names one (Railway: project, environment, service, deployment, volume). `controllerStamp` passes it to `proveOwnerGone`. A container redeploy changes the PID space, and the old stamp would read `different_process_space` (unprovable) forever. Now the existing `deployment_replaced` proof shows the old holder gone. This also covers `environment_run_lease_active` and the other kinds built on `controllerStamp`. - `currentDeploymentIdentity` in `owner-proof.ts` now delegates to `currentControllerDeployment`, so there is one parser. - Data doc: added the final Daytona wave, the six Daytona-only crash points (4 pass, 2 not reached, no violations), and the cause of the roughly 450 s codex TTFT. Runs waited behind the agent's harness verification, which restarted after each fault; boot itself took about 32 s. - Restart-chaos README: corrected which Daytona crash points a plain task reaches. ## Verification - Red, before the change: `environment-login-probe-lease-ownership.test.ts` failed 3 of 6. No settle happened, and the idle sweep released 3 live holders' leases. - Green: `pnpm exec vitest run` passes for `environment-login-probe-lease-ownership`, `environment-run-lease-ownership` and `heartbeat-orphaned-active-lease-sweep` (45/45), and for `environment-lease-owner-stamps`, `heartbeat-observed-workspace-preservation`, `ownership-reconciliation-report`, `heartbeat/recovery` and `workspace-recovery-warm-handoff`. - Redeploy: a stamp from a replaced Railway deployment settles with `deployment_replaced`. A stamp from the same deployment in another PID space stays `active`. This was red when `controllerStamp` dropped the deployment. - Live-owner counter-cases: the current process, another live PID, another PID space, and a lease re-stamped after the proof all stay `active`. A read-only pass never settles. - Not re-run live: the Daytona wave and crash points ran before this fix. The next Daytona wave should show no `environment_login_probe_lease` I2. ## Risks - A stamped run-less lease whose live holder never releases it, because of a bug in that process, is no longer released after five idle minutes. It is released once that process exits. This is the intended trade: never take a live holder's sandbox. - On an ephemeral server without a platform deployment identity (not Railway), a stamp from a replaced container is still unprovable. Such a run-less lease stays `active` until the provider expires the sandbox, and the reconciler reports it as unprovable. Before this PR, the five-minute sweep released it by age. This is the same open gap as unstamped legacy leases on an ephemeral server: there is no operator action yet. It is listed as a Phase 2 follow-up. - Legacy unstamped run-less leases keep the age fallback. These are rows written before #19965, and no current code writes them. ## Model Used - Anthropic Claude Opus 5.5 (`claude-opus-5-5`), via Claude Code with tool use. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
68d5fb5a6e |
feat(chaos): check I6 and I8 per crash point; ready the Daytona points (#19984)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - The on-demand crash-point suite (#19946) kills the control plane at named durable write boundaries and checks the plan's invariants > - Its first full run measured I1–I4 and I7, but I5, I6 and I8 were reported as not measured > - #19960 (generation stamping) and #19863 (speed budgets) now record what I6 and I8 need in the run log > - The six Daytona-only points have not run yet, and three of them cannot be reached by a plain task > - This pull request checks I6 and I8 per point, names the I5 gap, and readies the Daytona points for the coming Daytona wave > - The benefit is that every invariant is either asserted or a named gap, and a Daytona run explains any point it cannot reach ## Linked Issues or Issue Description Refs #19831 (plan, Phase 5). Follows #19946 (the suite). Uses #19960 (event generations) and #19863 (speed budgets). **Problem** - The suite reported I5, I6 and I8 as "not measured". - A Daytona run would report three points as `not_reached` with no reason. - A Daytona run without its key or image failed only after a 3-minute boot. **Expected behavior** - I6 and I8 are checked at each point. - I5 is a named gap with its reason. - Daytona points say what they need, and a Daytona run checks its inputs before it boots. ## What Changed - `tests/restart-chaos/invariants.ts`: `runLogViolations` checks two invariants from the run log, scoped to the wave's company: - I6: a run that failed with an error that names a stale or earlier-generation event; - I8: each `run.performance.budget_exceeded` event, with the budget evidence (metric, measured, p95, adapter, start kind). Both are keyed by run or event id, so the per-point baseline diff charges each one to the point that produced it. I5 is reported as `notMeasured` with a reason (`I5_GAP`). - `packages/shared/src/crash-point-registry.ts`: an optional `prerequisite` on the three Daytona points a plain task cannot reach. These are the teardown claim (needs a failed release), the reusable destroy (needs a `reuse_by_environment` lease), and stop-and-retain. - `tests/restart-chaos/crash-points.ts`: - `--list` prints each point's `needs:`, and a `not_reached` result names it too; - `--env daytona` refuses to start without `DAYTONA_API_KEY` and an image. - Docs: the README and the plan describe the I5/I6/I8 coverage and the Daytona points. ## Verification - `pnpm exec vitest run --config tests/restart-chaos/vitest.config.ts tests/restart-chaos/crash-points.test.ts`: 12 passed. - `cd packages/shared && npx vitest run src/crash-points.test.ts`: 15 passed. - `node scripts/typecheck-test-projects.mjs tests/restart-chaos` and `npx tsc --noEmit -p packages/shared` both pass. - I ran `runLogViolations` against a throwaway Postgres 18 with seeded rows, then stopped and deleted it. It returned: - I8 for the budget breach, with its evidence; - I6 for the run of this company that failed on `stale_generation_event`. It excluded the other company's stale run, the `process_lost` run and a non-budget event. - `node scripts/crash-points.mjs --list --env daytona` lists 19 points, with `needs:` on the three. `--env daytona` with no key or image fails at once with the missing names. - No harness run: the harness slot is held for other work. The next suite run exercises the new checks. ## Risks - Test harness only; no product code. - I6 is a proxy: it matches the run's error text and code. A failure that names a generation for another reason would be reported, with its evidence, for a person to judge. - I8 can flag breaches on a loaded host. Each breach carries its measured and budget values, so a load-caused breach is visible as one. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, via Claude Code with tool use (shell, file edits). ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
46bf913fb9 |
docs(plans): record the Pi steer duplicate after a restart and its follow-up (#19982)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - A person can steer a running Codex or Pi turn, and since #19970 a queued follow-up steers automatically > - Restart recovery (#19962) delivers a steered message that the model did not read before the restart, exactly once for Codex > - Pi gives the run log no read evidence for a steer, so after a restart a Pi steer that the model already read can arrive again > - The fix needs changes to the closure-pinned Pi distribution, which waits for the next Pi re-qualification > - This pull request records the limit and the follow-up plan so they are not lost > - The benefit is that the next Pi re-qualification can include the definitive fix ## Linked Issues or Issue Description **What is the documentation problem?** The Pi steering limit after a restart, and its planned fix, were only in pull request text and session notes. **Where?** `doc/plans/2026-11-07-pi-steer-restart-limit.md` (new). **What should it say?** Codex delivers an unread steered message exactly once after a restart. Pi can duplicate a message it already read, and never loses one. The note also gives the reasons, the follow-up plan (option A), its costs, and the rejected options. Refs #19962, Refs #19970 ## What Changed - New `doc/plans/2026-11-07-pi-steer-restart-limit.md`. It covers the limit, why Pi has no read evidence, the follow-up (forward Pi's native user `message_start` through pi-acp, acpx, the sidecar and Rust into a `userMessage` item), its re-qualification costs, and the rejected options. ## Verification - Documentation only. Read the note. The facts match `packages/paperclip-runner/src/steering-delivery.ts`, `patches/pi-acp@0.0.33.patch`, `pi-closure-pins.ts` and `rules/acpx-profile-identity.mdc`. ## Risks - None. Documentation only. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, via Claude Code with tool use. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
d33b3ac263 |
fix(wake-queue): save temporarily refused work intents and promote them through one path (#19956)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Users send work to agents as assignments, comments and Retries. The wake queue delivers each one as a run > - When an admission check refused a wake for a temporary reason, the wake often ended as a terminal skip, so the work was lost > - Restart-chaos waves found dropped comments and Retries (RC-37, s7532) and double delivery (RC-53). Each fix covered one hold type > - The durable ownership plan (section E) asks for one outbox rule: a closed set of terminal outcomes, a saved deferral for every temporary refusal, and one promotion path > - This pull request adds that rule. It saves messages refused by a pause, promotes deferred intents from every hold-release event through one function, and adds a reconciler backstop > - The benefit is that an accepted message is delivered exactly once after a pause lifts, and the reconciler reports any intent that is stuck or dropped ## Linked Issues or Issue Description Refs #19831 (plan, section E), #19838 (inventory, wake-queue rows and section 1b), #19927 (Phase 2 data: the I4 false positive), #19702, #19741, #19829, #19826. Builds on #19944 (section D, merged). **What happened?** - A user commented while the agent was paused, or while the company was paused. Admission refused the wake. Nothing saved the message as pending, so when the pause lifted, nothing delivered it. - Hold-release events woke different subsets of workers. A saved intent then waited out a 30-second refusal backoff after its hold cleared. - The I4 reconciler check matched free text with a regular expression. Phase 2 found a false positive: a delivered retry whose error said "retrying once". **Expected behavior** An accepted work intent ends only as delivered, consumed by a later run, task cancelled, or hard stop. Every temporary refusal saves the intent as `deferred` with its reason and its release trigger. Every hold-release event promotes deferred intents through one function (invariant 4). **Steps to reproduce** Pause an agent. Post a user comment on its task. Resume the agent. On `master` the comment is never delivered as a wake. ## What Changed - New `server/src/services/work-intent-outbox.ts`: - `WORK_INTENT_TERMINAL_OUTCOMES`: the closed set of terminal outcomes. - `TEMPORARY_REFUSAL_TRIGGERS`: maps each temporary refusal code to its release trigger. - `promoteDeferredWorkIntents`: the one promotion function. It removes the 30-second refusal backoff from the matching deferred intents. It then wakes run dispatch, the saved-receipt worker and stranded-work recovery, and hints each saved intent of the issue. - `heartbeat/queue.ts`: the #19829 save-as-deferred block is now a shared helper. A user's message refused by a paused agent, a paused company, or a budget pause that is not a reached hard stop is saved with `executionWait: { reason, trigger, message }`. The caller still gets a 409. The message is saved only when the task's latest run is terminal, because the release drain promotes a saved message behind that run. A task that never ran keeps the message in the task, and stranded-work recovery starts the task once the pause lifts. A `pending_approval` agent and a task-tree hold are not changed here: the release drain still fails a promoted wake for a pending approval, and tree holds have their own resume path. - Hold-release events call `promoteDeferredWorkIntents`: - accounting settled (`run-cost-accounting.ts`); - environment lease released (`environments.ts`, two sites); - native run projected terminal (`native-run-projection.ts`); - budget pause lifted (`budgets.ts`); - company reactivated (`companies.ts`); - task-tree hold released (`issue-tree-control.ts`); - agent lifecycle ready (`agent-lifecycle-driver.ts`). - Reconciler (`recovery/ownership-registry.ts`, `ownership-invariant-report.ts`, `native-finalization-reconciler.ts`): - registers three new kinds: `wake_request_deferred`, `queued_run` and `scheduled_retry_run`; - I4 now reads the closed refusal code (the wake `reason`) of accepted intents, not free text; - a new I4 check reports a deferred pause refusal whose pause has lifted for more than 5 minutes. The worker then promotes it as a backstop; - a lifted pause inside its 5-minute grace window returns its deadline through `nextDueAt`, and the worker also wakes on agent-lifecycle signals (company and agent pause changes). So a missed release event is still checked. - `doc/SPEC-implementation.md`: adds the rule "Accepted work intents are delivered exactly once". ## Verification Each focused file ran alone in one vitest process (`--maxWorkers=1`) on embedded Postgres. Results at `2875c85fd2`: | Question | Loop | Elapsed | Result | |---|---|---|---| | Is a comment refused by a pause saved, then delivered once after the pause lifts? | `heartbeat-held-comment-wake.test.ts -t "pause lifts"` | 8 s | 2 pass. With `queue.ts` from `master`, both fail: no intent is saved | | Does I4 use the closed code, and does it catch a stuck deferral? | `ownership-reconciliation-report.test.ts` | 45 s | 20 pass. This includes the Phase 2 "retrying once" false positive and the case "paused, then lifted" | | No regression on the paths that changed | held-comment (15), archived-company guard (8), undispatched execution wait (32), deferred receipt work (45), recovery pause durability (16), native run projection work (55), tree-control routes and services (19, 1, 7) | 8–334 s each | All pass | - Exactly-once: each pause test runs two promotions (a release event, then the reconciler backstop) and many dispatch passes. It asserts one delivery. The existing consumption tests (RC-53) still pass. - Server package typecheck passes. - Speed: nothing new runs before `turn.start`. The new admission work happens only on a refusal. Promotion runs in the release transaction and in the background reconciler. - Not run here: chaos waves. The coordinating session schedules them on the shared harness. ## Risks - `promoteDeferredWorkIntents` moves `updated_at` back for deferred intents that were refused in the last 30 seconds. A busy company with many release events can retry refused intents more often. Each retry re-runs normal admission, and the refusal resets the backoff. - More release sites now wake run dispatch and stranded-work recovery. These notifications are coalesced per company. - Lifting a pause is now the operator's explicit authorization to send what the pause refused, as Send is after a Stop. When the pause stopped the task's run, the lifted pause issues the task's next explicit wake, which adopts the saved message once. Other messages queued behind an acknowledged Stop keep the existing rule. If Stop recovery still holds the task, the message waits for that recovery. - An agent pause's stop sweep keeps the messages that the pause deferred (`preservePauseDeferred`). Terminate and cleanup stops still end them. - The backstop check reads only pause refusals (`company.inactive`, `agent.not_invokable`), because those holds can be checked exactly. Other deferred kinds keep their existing promotion paths. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, with tool use and code execution in Claude Code. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
051834bdc8 |
feat(chaos): add the on-demand crash-point suite and its manual CI workflow (#19946)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - A control-plane restart between two durable writes leaves an in-progress record owned by a dead process > - Random restart-chaos waves found these 20 ms–1 s windows only by luck > - #19937 added `crashPoint(name)`, and #19938 put it at 20 durable write boundaries > - Nothing yet runs a task, kills the server at a chosen point, restarts it, and checks the plan's invariants > - This pull request adds that on-demand suite, as a local command and as a manual CI workflow (not a required check, per Decision 2) > - The benefit is a repeatable work list: which boundary breaks which invariant, for Phases 2–4 ## Linked Issues or Issue Description Refs #19831 (plan, Phase 5 and Decision 2). Refs #19861 (report-only reconciler used for I1–I7). Stacked on #19938 (call sites), which is stacked on #19937 (helper). **Problem** The named crash points cannot be exercised end to end. No command arms one point, runs real agent tasks across it, restarts, and grades the result. **Expected behavior** `node scripts/crash-points.mjs --points <glob> --kinds <id> --adapters <list>` runs one wave per selected point and reports each point as pass, fail (with failing R rules and violated invariants), not reached, or harness error. A `workflow_dispatch` workflow runs the same thing on demand. ## What Changed - **Runtime arming.** `PAPERCLIP_CRASH_POINT_CONTROL=<file>`, in the shared helper and the runner mirror. Each hit reads `<token> <name>[:<nth>]`, and a new token resets the count. It is test mode only: `NODE_ENV=production` ignores it, and the disarmed path is still one comparison. - **`tests/restart-chaos/main.ts --crash-point <a>:<nth>,<b>:<nth>,...`** (config, types, ledger): - one booted server and no Vite UI; - per point: take an invariant baseline, launch the point's batch, arm the point, then on rest disarm it and check invariants as violations that are new since the baseline; - every task is graded R1–R6 and T; - `crash-point.json` (v2) holds one entry per point. - **`tests/restart-chaos/supervisor.ts`:** - every spawn gets the control file; - a SIGKILL whose receipt names the child's pid is restarted like K1: the control file is emptied first, the owned tree is killed, and a replacement starts. A failed tree cleanup fails the fault; - the isolated instance pins `PAPERCLIP_IN_WORKTREE=false` (a `.paperclip/.env` above the checkout otherwise suppressed all scheduling); - `RESTART_CHAOS_HEALTH_TIMEOUT_MS`. - **`tests/restart-chaos/invariants.ts`:** - the Phase 1 report-only pass with `record: false`: two real-clock passes for I1 and I2, and one pass 15 minutes ahead for I3, I4 and I7; - spool kinds are read from the wave's instance; - an unfinished pass is a harness error; - each violation carries its wake and run state as evidence. - **`tests/restart-chaos/crash-points.ts` and `scripts/crash-points.mjs`:** - select points by glob, kind and environment, and run them in one wave; - a per-point outcome: pass, fail, not_reached or harness_error; - `SUMMARY.md` with a violation evidence column. - **`.github/workflows/crash-points.yml`**, manual only, under the paid-workflow contract: - a credential-free `authorize` job; - a `suite` job in `runner-e2e-paid` whose first step reauthorizes; - a frozen install with `--ignore-scripts` and no shared cache; - only the selected adapters' keys; - the workflow is added to both security-test allowlists. - **Docs:** "Crash-point suite" in `tests/restart-chaos/README.md`, and "Built (Phase 5)" plus the first work list in the plan. ## Verification - Focused tests: - `pnpm exec vitest run --config tests/restart-chaos/vitest.config.ts tests/restart-chaos/crash-points.test.ts`; - `cd packages/shared && npx vitest run src/crash-points.test.ts`: 15 tests, including runtime arming in a real child process; - `node --test scripts/__tests__/release-verify-workflow.test.mjs`: 72 tests; - `tests/runner-e2e/workflow-security.test.ts`: 19 tests. - Typecheck: `node scripts/typecheck-test-projects.mjs tests/restart-chaos` passes. - Loop line: question "how long is one crash-point loop?" | loop: `node scripts/crash-points.mjs --points <one>` (codex A1) | elapsed 2.7 min end to end: boot 9 s, crash 9 s after arming, replacement ready in 6 s, rest about 2 min | result: one point alone takes about 3 min. Before this change it took about 15 min (one cold boot per point, Vite UI, host at load 77). - Full local run, 14 points, codex A1, 11.8 min in all: - every point was reached (the server killed itself at each one, and the replacement came back); - 11 pass; - `workspace.finalize.after_start_before_copyback` and `workspace.finalize.after_copyback_before_done` fail I2 and R6: a `workspace_finalize` operation stays running after the run succeeded; - `wake.claim.after_run_running_before_wakeup_claimed` reports I4. Its evidence shows the wake failed with "…retrying once", and the retry delivered the work, so this is the known refusal-regex false positive. - The 6 Daytona-only points were not run. ## Risks - Test harness and an on-demand workflow only. No product code changes. - `.github/**` needs a CODEOWNER review. The workflow uses provider keys, so it refuses non-default branches and actors that are not on the allowlist, like `runner-full-stack-e2e.yml`. - `PAPERCLIP_IN_WORKTREE=false` is pinned in every chaos wave, not only crash-point waves. A chaos instance is never a worktree dev instance, so scheduling must not be suppressed. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, via Claude Code with tool use (shell, file edits). ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Devin FoleyandPaperclip
|
ab82ac0191 |
fix(workspaces): retire recovered workspace copies after a successful turn (#19979)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - With remote (sandbox) execution, every turn ends by saving the whole working folder, so the next turn can continue on any sandbox. > - When a run is cut off mid-turn and its task continues in a new workspace, preparation attaches the dead run's saved workspace under `.paperclip-recovered/<operationId>`, so the agent can recover unfinished work. > - Nothing ever removes that attachment. Every later save archives and uploads it again, and a new sandbox restores it again. > - The attachment is a full workspace, `.git` included, so every save for that task is roughly twice as big. The extra size lasts until the task ends. > - This pull request removes an attachment after a run that was shown it succeeds, and keeps a retired record so it is never attached or shown again. > - The benefit is that affected tasks go back to normal save sizes after their first successful turn, without changing how recovered work reaches the agent. ## Linked Issues or Issue Description **What existing behavior does this improve?** Recovery attachments in remote workspaces (`.paperclip-recovered/<operationId>`). **Subsystem affected** Server: remote workspace preparation and recovery attachments. **Current behavior** `retainRemoteRecoveryAttachments` keeps every recorded attachment on every later preparation, and no code path removes one. On one hosted instance, 10 active workspaces carried 18 attachments. One task's saves went from 746 MB to 1,472 MB at a single save and stayed there. The largest grew from 709 MB to 3.3 GB. A 3.3 GB workspace needs most of a 10 GB sandbox disk during each save. When a task moves again, the new attachment contains the older one, so the copies nest. **Proposed behavior** Each live attachment records the run it was shown to. When that run has succeeded, the next preparation removes the attachment from the sandbox, after checking its exact provenance. It keeps a `retired` record, so: - a sandbox restored from an older save is cleaned up too; - the attachment is never attached or shown to an agent again. **Reason and benefit** A run that saw the attachment and succeeded had a full turn to use it. Removing it then returns the task's saves to normal size and frees sandbox disk. **Breaking changes** None. Records written before this change have no `presentedToRunId`. They are stamped on their next preparation and retired after the following successful run. Related: none found. ## What Changed - `remote-workspace-recovery-attachments.ts`: - New `retiredRecoveryAttachments`. It returns attachments already retired, and attachments shown to a different run that has since succeeded. - `retainRemoteRecoveryAttachments` removes those attachments from the sandbox and keeps a `retired: true` record for each. - `attachPreservedRemoteWorkspaceSnapshots` skips retired operation IDs. - `workspace-recovery-snapshot.ts`: `attachPreservedWorkspaceSnapshots` also skips retired operation IDs. - `remote-workspace-transport.ts`: new `removeRemoteRecoveryAttachment`. - It reuses the attachment probe's checks: a real directory under the workspace root, and an exact `provenance.json`. - It makes read-only directories writable before removing them. - It removes `.paperclip-recovered` once it is empty. - Anything else at that path fails the preparation instead of being deleted. - `remote-workspaces.ts`: computes the retired set before attaching. It writes `presentedToRunId` on each live record, and keeps retired records unchanged. - `heartbeat/workspace-preparation.ts`: never shows retired records to an agent. - `doc/execution-semantics.md`: describes retirement. ## Verification - `pnpm --filter @paperclipai/server typecheck`: exit 0. - New tests: - `remote-workspace-transport.test.ts`: removal against real sandbox commands. It removes an attachment with exact provenance, including a read-only directory, and leaves sibling attachments. It refuses a foreign proof or a symlinked directory without deleting anything. It reports a missing attachment, and removes the empty parent folder. - `workspace-recovery-preparation.test.ts`, new `retired_followup` case. The run shown the attachment succeeds. The next preparation, on a fresh sandbox restored from that run's save, removes the attachment and records it as retired. A later preparation after a failed run still neither attaches nor shows it. - `ordinary_followup` (the agent deleted the attachment itself) now also ends with a retired record instead of an empty list. - A retired operation ID is not attached again, even to an empty destination. - 13 recovery and remote-workspace suites: 836 passed. With the source change reverted, the 3 new or changed preparation cases and the 5 new transport cases fail. ## Risks - **An agent can lose access to recovered work after one successful turn.** That happens if a successful run never looked at the attachment. The source run's saved snapshot in storage is not touched, so an operator can still recover it. - **Retired records stay in workspace metadata.** There is one small record per attachment, and each later preparation sends one sandbox probe per record. - **Nested attachments are unchanged.** If no turn succeeds between two moves, the newer attachment still contains the older one. That older copy may hold work nobody has recovered yet, so this change does not drop it. - Low risk overall: the change is limited to remote workspace preparation and is fail-closed on provenance. > For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and discuss it in `#dev` before opening the PR. Feature PRs that overlap with planned core work may need to be redirected — check the roadmap first. See `CONTRIBUTING.md`. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`) in Claude Code, with tool use (shell, file edits, test runs, read-only database queries). AI-assisted investigation, change and tests, with human direction. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge --------- Co-authored-by: Paperclip <noreply@paperclip.ing> |
||
|
DottaandClaude Opus 5.5
|
f6cb3e43f4 |
fix(admission): admit new work independent of other runs' cleanup and accounting (#19944)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Budget admission decides if new agent work can start in a company > - Five admission gates waited on the cleanup or accounting of other runs > - The restart-chaos campaign found that these gates froze companies for up to 10 minutes, caused false budget pauses, denied in-run writes, and dropped wakes, comments and Retries > - The durable ownership plan (Decision 1) removes the company-wide accounting gate. It does not narrow it again > - This pull request removes the five gates. Unfinished accounting now adds its bounded estimate to exposure > - The benefit is that one run's unsettled accounting never stops the other work in its company ## Linked Issues or Issue Description Refs #19831 (plan, section D), #19838 (inventory, finding 3), #19927 (Phase 2 reconciler data), #19497, #19724, #19826. **What happened?** Admission refused new work in these cases, with no reached hard stop and no pause: - a terminal run awaited accounting (`waitingRunCount`); - a native run recovered its accounting (`recoveringRunCount`); - a budget-owned scope hold awaited reconciliation, after the hard stop no longer applied; - a budget input was older than 15 s (`budget_evaluation_work`); - the predecessor run row of a task was missing (`hasPendingRunCleanup` failed closed). **Expected behavior** Admission never waits on the cleanup or accounting of another run. Only a reached hard stop, an explicit pause or an inactive company refuses new work (invariant 3). **Steps to reproduce** Make a company with an active budget policy. Leave one failed run with `cost_accounting_pending = true`. Request work for another agent. On `master` admission refuses with "Company cannot start work while completed runs await accounting." ## What Changed - `budgets.ts`: admission refuses only for a reached hard stop or an explicit pause. The code removes `observedBlocks`, the `recoveringRunCount` check, the stale `budgetHeld` refusal and the 15 s input-age gate. It also removes the unused `inspectInvocationEligibility`. `reconcileScope` no longer returns accounting holds. - `budget-exposure.ts`: removes the `waiting` and `recovering` aggregates and the own-task subquery. The admission query is smaller. Unfinished accounting stays exposure: the larger of the held reservation and the observed cost. - `budget-evaluation-work.ts`: removes `oldestBudgetInputAt`, which nothing reads now. - `execution-blocker.ts`: `hasPendingRunCleanup` returns false for a missing run row. A run row that is deleted removes its finalization coordinator and nulls its lease references, so the row owns nothing. A predecessor that owns a lease or a coordinator still holds the successor of its own task. - `recovery/ownership-invariant-report.ts`: the I3 check no longer mirrors the removed gate. It reports a wake request or a run that was refused with an accounting-hold reason. Any such report is a regression. - `doc/SPEC-implementation.md`: adds the product rule "Admission is independent of cleanup and accounting". `doc/execution-semantics.md` is aligned. ## Verification - New `server/src/__tests__/admission-independent-of-cleanup.test.ts`, on embedded Postgres. It has one case per gate. All six cases fail on `master` and pass here. A seventh case (an explicit pause still refuses) passes on both. One case also asserts invariant 3 together with exposure: two failed runs that each hold a 400-cent reservation fill a 1000-cent budget, and a hard stop still refuses. - Existing tests that asserted the old gates now assert admission (invariants 3 and 4): `closed-native-receipt-accounting`, `budgets-service`, `budget-evaluation-work`, `heartbeat-held-comment-wake`, `ownership-reconciliation-report`. In `heartbeat-held-comment-wake`, each saved comment, Retry, assignment and continuation is delivered exactly once while another run's accounting is pending. Settling accounting later does not deliver it again. - All six files pass locally, one file per process. The server package typecheck passes. - Speed: admission is on the turn path. This change only removes work from it: two correlated aggregates, one own-task subquery and one `budget_evaluation_work` read per policy. No wait or I/O is added before `turn.start`. - Not run here: the chaos waves (K13 storm, K6 crash loop, K4 long holds). The coordinating session schedules them on the shared harness. ## Risks - Behavior change: work starts while accounting for other runs settles. Spend stays bounded by the reservation capacity check and the hard stop, which read the ledger plus observed cost. A run that has no reservation and no observed cost adds zero exposure until its charge commits. This is the gap that Decision 1 accepts. Budget enforcement was already best effort (SPEC 11.6). - A budget-owned pause with no reached hard stop no longer refuses work in budget admission. The paused company or agent is still refused by the existing pause checks until the budget worker lifts the pause. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, with tool use and code execution in Claude Code. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
a4a3e342ec |
fix(recovery): charge a restart attempt only when the provider accepted a turn and made no progress (#19948)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - A run has an execution recovery budget: three provider attempts for native runs, and two failure retries for legacy runs > - In the restart chaos campaign, control-plane restarts used up that budget for runs that were healthy (RC-22, 22b, 22c, 40 and 60) > - #19720 and #19747 fixed two variants with watermark heuristics, and legacy interrupted preparation still spent the failure budget > - The durable-ownership plan (section F, invariant 5) asks for one rule: charge an attempt only when a recorded outcome proves that provider work started and no progress followed > - This pull request adds that rule and moves the native restart charge and the legacy preparation retry onto it > - The benefit is that a restart never spends budget for work the provider never started, while real failure loops and crash loops stay bounded ## Linked Issues or Issue Description Refs #19720, #19747, #19856 **What happened?** A control-plane restart charges a provider attempt by heuristics. The first crash restart always charges. A later restart charges unless the previous resume had no start evidence (#19747) or completed work (#19720). A runner that spawned but whose provider never accepted a turn counted as started, so a K6 crash loop could still spend the budget. Legacy runs whose preparation a restart interrupted (`providerWorkStarted: false`) spent `failureRetries`, the same counter as real failures (`execution-recovery-attempt.ts`). **Expected behavior** An attempt is charged only when the provider accepted a turn (`turn.accepted` or `turn.started`) and no accepted turn then completed work. A startup before spawn, a runner that spawned but was never admitted, and a `turn.start` the provider did not accept are not charged. The generation-12 cap still bounds uncharged restarts. **Steps to reproduce** Start a native run at its last attempt. Kill the controller after the runner spawns but before the provider accepts a turn. Before this change, the next restart claim blocks the run with `execution_recovery_budget_exhausted`. ## What Changed - New `server/src/services/native-runtime/provider-attempt-outcome.ts`: the one rule. `readProviderAttemptOutcome()` returns `provider_not_started`, `provider_started_without_progress` or `provider_progressed` from run events. `providerAttemptChargesBudget()` charges only the second. `providerAttemptWindowStart()` starts the attempt window at the latest `native.process_start_requested`, so earlier work never counts for a later attempt. `PROVIDER_ATTEMPT_CREDIT_GENERATION_LIMIT` (12) is the shared cap. - `restartResumeKeepsAttempt()` (`native-restart-recovery.ts`) now applies that rule to the attempt the restart interrupted, including the first restart. It keeps the unstarted-resume carry from #19747, the reattachment rule (an adopted provider counts as started), the drain exemption from #19625, and the generation-12 cap. A history entry from an older release without a recorded position still charges. - Legacy runs: `interrupted_preparation` is now an uncharged retry lane (`execution-recovery-attempt.ts`). It keeps its own counter, bounded at the same cap (12), and never changes `failureRetries`. `retries.ts` uses the longest transient delay after the second lane attempt. - `doc/execution-semantics.md` states the rule. - Accounting has its own "provider work started" readers (inventory Summary finding 5). Phase 3 owns those; this PR exposes the rule for them to adopt and does not change them. ## Verification - `native-runner-restart-recovery.integration.test.ts` (embedded Postgres, real `claimNativeRestartRecoveries`): 129 passed. New cases, red on master: - not charged at the last attempt for a claimed remote bootstrap, a remote bootstrap retry, and a claimed unlaunched remote run; - not charged when the runner spawned but was never admitted, and when the provider did not accept `turn.start`; - not charged on the first restart when the original attempt already made progress; - a normal retry that spawned but was never admitted is not charged, and earlier work still does not count for it. - Counter-cases, green on master and with the change: an accepted `turn.accepted` or `turn.started` with no progress is charged; the generation cap blocks at the last attempt; crash loops with and without progress stay bounded; the #19720 and #19747 cases still pass. - `execution-recovery-attempt.test.ts`: new lane tests (red on master). - `workspace-recovery-preparation-dispatch.test.ts` (25 passed), `heartbeat-process-recovery.test.ts` (interrupted-run cases), `run-dispatch/adapters/postgres.test.ts` (interrupted cases). - `server` typecheck (`scripts/typecheck-layers.mjs`, low-memory mode): no errors. - Not run: chaos waves (the coordinator schedules them on this host), and the full server suite (CI covers it). ## Risks - Behavior change: the first crash restart no longer charges when no accepted turn is recorded. If a crash lands before the controller projects the provider's `turn.accepted`, that attempt is not charged. This fails toward more retries, and the generation-12 cap bounds it. - A legacy preparation crash loop now gets up to 12 retries instead of 2. Each retry still passes the normal issue gates and keeps the transient delay. - The new queries use the `(run_id, seq)` index, once per restart claim. Nothing new runs before `turn.start`. > For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and discuss it in `#dev` before opening the PR. Feature PRs that overlap with planned core work may need to be redirected — check the roadmap first. See `CONTRIBUTING.md`. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, via Claude Code with tool use (shell, file edits, test execution). Extended reasoning enabled. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
b4842df1e4 |
docs(plans): Phase 2 Daytona wave results (#19947)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - Phase 2 of the durable-ownership plan moves in-progress record kinds to one shared ownership path, guided by chaos data. > - The first data doc (#19927) covered local waves only. Remote runners, leases and sandboxes need a Daytona wave. > - This pull request adds the results of one Daytona wave to the data doc. > - The benefit is that Phase 2 and Phase 3 prioritize remote kinds from measured data. ## Linked Issues or Issue Description Refs #19927 (data doc), #19934 (claimed-wake settle), #19863 (speed budgets). **What is the problem?** The Phase 2 data doc had no remote data. **Where is it documented?** `doc/plans/2026-11-07-phase2-reconciler-data.md`. **What should change?** Add the Daytona wave's per-kind violations, settle latency, failures and speed numbers. ## What Changed - Added a "Daytona wave" section: 6 tasks with K1, K2 and two K4 outages of about 575 s. 5 of 6 passed, and R6 was clean. - Recorded the t03 failure: a Retry was cancelled by the company-wide accounting gate (Phase 3 scope). - Recorded the claimed-wake leak again (#19934 fixes it) and remote settle times of 76–107 s. - Corrected the earlier note about the image content-ID mismatch. ## Verification - Wave `w-20261108T012041-s9301`; the numbers come from a copy of the kept database and the read-only ownership sampler. - This is a docs-only change. ## Risks - Low. Docs only. The samples are small and the doc says so. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, in Claude Code with tool use (shell, file edits). ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [ ] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
f35eb1e5ef |
fix(observability): judge the TTFT budget from run creation; report the wait behind a live turn (#19941)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - Invariant 8 of the durable-ownership plan says turns must start fast. #19863 added per-run speed budgets that write `run.performance.budget_exceeded` to the run log. > - The Phase 2 chaos data (#19927) showed follow-up TTFT about 3x over the claude budget. > - The cause is measurement scope. A follow-up comment posted during a live turn waits for that turn to end, and TTFT is anchored at the comment. The budgets came from idle follow-ups. > - This pull request judges the TTFT budget from run creation and reports the wait separately. > - The benefit is that budget breaches point at real start-up slowness, not at time spent behind a turn the user is already watching. ## Linked Issues or Issue Description Refs #19863 (speed budgets), #19927 (Phase 2 data), #19852 (startup sub-spans), #19888 (warm default). **What happened?** In two local storm waves, C1 follow-up runs breached the claude warm TTFT budget (p50 27.6–29.9 s against 9.5 s). Their spans show `comment.to_run_created` at 6–62 s (waiting for the live turn), then session resume at 1.4–3.8 s and submit-to-first-event at 2–5 s. Codex without the wait was about 4.8 s, the same as the warm-follow-up baseline. **What did you expect?** The budget should judge start-up latency, which is what its baseline measured. The wait behind a live turn should be visible but not judged. **Steps to reproduce** 1. Run a local chaos storm with C1 tasks. 2. Read `run.performance.budget_exceeded` rows with metric `ttft` for `issue_commented` runs, and compare them with each run's `comment.to_run_created` span. ## What Changed - `native-run-trace.ts`: the TTFT budget check measures from run creation (start of `heartbeat.queue`, or end of the ingress span). The TTFT span itself does not change. - `native-speed-budgets.ts`: breaches can carry `inputWaitMs` (TTFT only). - `tests/restart-chaos/ttft-baseline-stats.ts` and `ttft-baseline.ts`: the wait behind a live turn is its own measured step (`inputWait`), next to the full TTFT and the TTFT from run creation. The TTFT table gains a "Wait behind live turn" column. The full comment-to-first-token time stays the user-visible metric. - `doc/run-log-events.md` and the plan's speed-budget section describe the rule. ## Verification - `cd server && npx vitest run src/services/native-runtime/native-speed-budgets.test.ts src/services/native-runtime/native-run-trace.test.ts`: 43 pass, 2 new. `pnpm exec vitest run --config tests/restart-chaos/vitest.config.ts tests/restart-chaos/ttft-baseline-stats.test.ts`: 12 pass, 1 new. They cover a follow-up under budget after a 60 s wait (no breach), a follow-up over budget (breach with `inputWaitMs`), and queueing after run creation still judged. - Server typecheck passes. ## Risks - Low. This is a run-log-only change (AGENTS.md §5.7): no Telemetry and no span attributes change. - A slow wait before run creation is no longer a TTFT budget breach. The TTFT span still records it, and the breach payload names it. Admission holds after run creation are still judged. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, in Claude Code with tool use (shell, file edits). ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [ ] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
17b9e10252 |
docs(plans): Phase 2 reconciler data; let chaos K4 holds reach 600 s (#19927)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - Control-plane restarts leave in-progress records that each subsystem releases in its own way. The durable-ownership plan replaces those paths with one shared path. > - Phase 1 added a report-only reconciler. Its light local waves showed zero violations, and that proves little. > - Phase 2 moves kinds to the shared path. It needs data that shows which kinds really leak, and the speed numbers to protect. > - This pull request records that data from harsher waves. It also fixes the harness so that K4 outages past the reconnect grace qualify. > - The benefit is that Phase 2 moves kinds in order of measured risk, against a known speed baseline. ## Linked Issues or Issue Description Refs #19831 (durable-ownership plan), #19861 (report-only reconciler), #19863 (speed budgets), #19797 (K4 hold cap). **What is the problem?** Phase 2 needs per-kind invariant data and speed data from harsh restart waves. Only light local waves existed. In addition, a K4 hold over about 405 s always failed its fault receipt, so the 600 s cap from #19797 could not be used. **Where is it documented?** `doc/plans/2026-11-07-durable-ownership-and-reconciliation.md` (Phase 2 rows) and `tests/restart-chaos/README.md`. **What should change?** Record the data in a dated plan doc. Let chaos K4 holds run up to 600 s and still qualify. ## What Changed - Added `doc/plans/2026-11-07-phase2-reconciler-data.md`. It covers 5 unmasked local waves (two storms, a crash loop, two K4 runs). It gives per-kind violation counts with example rows, owner verdicts, TTFT, boot and restart recovery against the budgets, Phase 3 gate notes, and what is not covered (Daytona). - Added a short pointer to that doc in the durable-ownership plan. - `tests/restart-chaos`: new flag `--k4-hold-max-seconds` (120–600). It raises the upper end of the seeded K4 hold. The default range (120–300 s) does not change. - `tests/runner-e2e/controller-outage.ts`: the receipt's ready bound is now a parameter (`CONTROLLER_READY_BY_MS`, default 420 s). Runner suites keep 420 s. The chaos supervisor passes hold + 420 s, so a hold past the Runner's reconnect grace qualifies. ## Verification - `pnpm exec vitest run --config tests/restart-chaos/vitest.config.ts`: 12 files, 135 tests pass, including the new flag test. - `pnpm exec vitest run --config tests/runner-e2e/vitest.config.ts tests/runner-e2e/controller-outage.test.ts`: 36 pass, including the new long-hold receipt test. - `node scripts/typecheck-test-projects.mjs tests/restart-chaos` and `tests/runner-e2e`: pass. - Live: the K4 wave at seed 9203 failed 3 of 6 tasks with the old bound (fault error "did not produce a qualifying replacement receipt"). It passed 6 of 6 with this fix, with holds of 294–579 s. ## Risks - Low risk. The changes are test harness code and docs only. The runner suites keep the 420 s ready bound by default. - The data is local only. The doc says which kinds the waves do not reach. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, in Claude Code with tool use (shell, file edits). The model ran the chaos waves and wrote the analysis and code. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [ ] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
7fd98682ca |
fix(workspaces): give every subtask its own execution workspace (#19860)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Agents split work into child tasks, and each task runs in an execution workspace (a checkout, branch, or sandbox folder) > - Since early 2026, a new child task used its parent's realized workspace by default: the same checkout, branch and sandbox > - A parent and its child often run at the same time. Two writers in one folder lose each other's files, the waking parent is cancelled with `workspace_busy`, and sandbox lease ownership crosses from the parent into the child, so retained-file recovery refuses the workspace as ambiguous > - The product rule is simple: a subtask gets its own workspace, independent of the parent's saved files. Paperclip never needs to carry a parent's unmerged changes into a child > - This pull request makes that the only behaviour on every creation path, and moves existing shared children to their own workspace on their next run, with no user or admin action > - The benefit is that parents and children never block or overwrite each other, and recovery ownership stays with one task ## Linked Issues or Issue Description **What happened?** A child task inherited its parent's realized execution workspace by default (`createChild` defaulted to `executionWorkspaceInheritanceMode: "linkage"`, and `issueService.create` fell back from `inheritExecutionWorkspaceFromIssueId` to `parentId`). When the parent woke while the child wrote to the shared workspace, the parent run was cancelled with `workspace_busy`. In sandbox environments, the lease ownership chain crossed from the parent into the child, and retained-file recovery refused the workspace with `allocation_ownership_ambiguous`. **Expected behavior** Every new child task gets its own execution workspace, realized from the project base exactly as a top-level task in that project. It never uses the parent's checkout, branch, saved snapshot or unmerged changes. A parent and its descendant never share a workspace by any route. **Steps to reproduce** 1. Run a parent task in a project with isolated (git worktree) or sandbox workspaces, so the parent has a realized execution workspace. 2. From the parent run, create a child task with only `parentId` (or through the runner `create_task` tool, chat suggestions, decision effects, plugins, or the New sub-task dialog). 3. See that the child's `executionWorkspaceId` is the parent's, with `executionWorkspacePreference: "reuse_existing"`. 4. Wake the parent while the child runs in a sandbox environment. The parent run is cancelled with `workspace_busy`. **Additional context** Refs #9233 (the existing `strategy_only` opt-out). The opt-out `executionWorkspaceInheritanceMode: "strategy_only"` already existed for accepted-plan decomposition. This change makes that behaviour (tightened to drop a pinned existing branch) the only behaviour. ## What Changed - **One rule for every creation path (`server/src/services/issues.ts`).** `issueService.create` still uses the parent (or `inheritExecutionWorkspaceFromIssueId`) as a source for the project, project workspace and workspace *strategy*, but it never copies the source's `executionWorkspaceId`, `reuse_existing` preference, or pinned existing branch. All server paths that create a task go through `create()`/`createChild()`: issue routes (`POST /issues`, `POST /issues/:id/children`, the run-bound default parent and run workspace source), runner `create_task`, chat suggestion acceptance, native status decisions, accepted-plan decomposition, decision effects, plugins (`ctx.issues.create`), external agents, watchdog follow-ups, monitor recovery issues, case handoffs and email channels. - **Explicit ancestor workspaces are dropped.** If a create names an `executionWorkspaceId` (or an `existing` workspace selection) that an ancestor of the new task created or is bound to, `create()` drops it and keeps only the strategy. Older clients and skills that copied the parent's id keep working and get an independent workspace. - **`createChild` no longer has a "linkage" mode.** `executionWorkspaceInheritanceMode` is still accepted from internal callers and is ignored. - **`inheritExecutionWorkspaceFromIssueId` decision.** I kept the field for API compatibility, but it now copies only the source issue's project and workspace strategy. I removed its workspace-sharing behaviour for every pair, not only parent/child: the source task and its follow-up can run at the same time (the runner, watchdog and status-decision callers all create the follow-up while the source run is live), so sharing would again let two tasks write one folder. Explicit sharing between unrelated tasks stays available through an explicit `existing` workspace selection. - **New `server/src/services/task-workspace-lineage.ts`.** `buildIndependentTaskWorkspaceSettings` (strategy only: mode, environment, runtime, network egress, shared-workspace concurrency, base ref, branch template, commands; no `reuse_existing`, no `operator_branch`, no `existingBranch`), `loadTaskLineageIssueIds` (the parent chain plus, for a task that depth placement put beside its requester, the requester's chain via `privacyParentIssueId`) and `isAncestorExecutionWorkspace` (the workspace's `sourceIssueId` is in that lineage, or a lineage issue is bound to it). A child in another project does not take the parent's strategy. - **Upgrade-safe rebind at admission.** `executionWorkspaceService.releaseAncestorWorkspaceBindingAtAdmission` runs at the same admission point as the pending workspace selection (`heartbeat.ts`). For a child (or a flattened task beside its requester) that an older release bound to an ancestor's workspace, it clears the child's binding (keeps the project workspace and strategy, bumps the binding revision) so the run realizes the child's own workspace. It does nothing while another run of the child is running, while the child's earlier run still owns workspace finalization or recovery, for native restart/lease-owner recovery, for a process-loss retry, or for a Retry that keeps the current workspace. The parent and its workspace are never changed. - **Projects without a project workspace (per-task repository acquisition).** A parent with no project workspace worked in its own task files, so its settings (for example `isolated_workspace`) do not describe the project base. A child of such a parent, and an older child released from one, takes the project's defaults exactly like a top-level task, instead of an isolated mode that would resolve to a git worktree with no git base ("needs a project / project workspace"). When the child's new workspace is created at admission, `executionWorkspaceRepositoryService.prepareForAdmission` repeats the parent's (or requester's) repository requests as the child's own pending requests (identity, URL, catalog id and requested ref only; no pinned commit, branch or files). The normal acquisition path then clones a fresh copy into the child's task files with the child's own access checks. This runs only for a workspace created by that admission, so existing tasks never gain repositories. - **UI.** The New sub-task dialog no longer pre-selects the parent's workspace. `buildSubIssueDefaults` drops `executionWorkspaceId` and `reuse_existing`, and derives the parent's independent mode (`shared_workspace`, `isolated_workspace` or `agent_default`) from its preference, saved settings or current workspace. The "This sub-task will no longer use the parent task workspace" warning is removed. - **Contracts and docs.** Shared validator and plugin SDK comments describe the new meaning of `inheritExecutionWorkspaceFromIssueId`. Updated `doc/execution-semantics.md` (new rule in §6 and a wording fix in the retained-sandbox paragraph), `doc/SPEC-implementation.md` (child workspace rule, requester keeps *strategy* inheritance, watchdog follow-ups), `doc/SPEC.md` (task workspace ownership), `docs/specs/external-task-protocol.md`, `skills/paperclip/SKILL.md` (step 9 and the workspace bullet), `skills/paperclip/references/issue-workspaces.md`, the CEO onboarding `HEARTBEAT.md`, and the plugin SDK README. ### Previous-release story (`rules/upgrade-safe-run-state.mdc`) - **Migrated on next admission.** A child that the previous release bound to its parent's (or any ancestor's) execution workspace keeps that binding until its next normal admission. Then the server clears the binding and the run realizes the child's own workspace from the project base. No schema migration is needed. No user or admin action is needed. - **In-flight writers finish first.** A running child is not interrupted. Its current run, its native restart recovery (admitted native input or lease owner), a process-loss retry, and a Retry that keeps the current workspace all finish in the shared workspace. While a child's earlier run is still running or its workspace finalization or recovery is unsettled, the binding stays and a later admission releases it. - **No files are dropped.** The release changes only the child's binding row. The parent keeps its binding, and the shared workspace row, folder, branch and saved snapshots are not touched. Anything the child left there stays in the parent's workspace. - **No new refusal codes.** Saved run identity (native execution input, leases, checkpoints) is not changed or refused; the rebind only affects new admissions. ## Verification - New `server/src/__tests__/subtask-workspace-independence.test.ts` (embedded Postgres): `create` with `parentId`, `createChild` with no mode / `"linkage"` / `"strategy_only"`, `inheritExecutionWorkspaceFromIssueId` pointing at the parent, explicit parent and grandparent `executionWorkspaceId`, a non-child follow-up, and unchanged top-level defaults and explicit top-level reuse. 7 creation cases fail on `master` and pass with this change. Admission cases: an old-release child is rebound while its parent is running and the parent keeps its workspace; a running child is not rebound (deploy lands mid-run); an unrelated explicit share is kept; the parent is never rebound. - New heartbeat case in `server/src/__tests__/heartbeat-workspace-busy.test.ts`: an old-release child bound to its running parent's workspace in a sandbox environment runs to `succeeded` with no `workspace_busy`, gets a new workspace, and the parent run and binding stay unchanged. It fails on `master` (the child waits with `workspace_busy`). - Updated tests that asserted the old inheritance in `issues-service.test.ts`, `task-workspace-binding.test.ts`, `chat-project-tools.test.ts`, `task-workspace-child-trust.test.ts`, `heartbeat-git-restore-point-resume.test.ts` (a child bound to a closed parent's expired workspace now starts from the project base and leaves the parent's restore point alone), and `ui/src/lib/subIssueDefaults.test.ts`. Remote-workspace reservation, continuation and writer-wait suites now seed the older-release shared shape directly (or an unrelated explicit share), so those safety paths stay covered. - `npx vitest run` on the changed suites and about 45 related server suites (heartbeat workspace, recovery admission, issue routes, routines, plugins), `ui` `subIssueDefaults` and `NewIssueDialog` tests, `pnpm check:token-gates`, `pnpm -r typecheck`. - New `server/src/__tests__/heartbeat-subtask-acquired-repository.test.ts`: a project with no project workspace, a parent whose task files hold an acquired repository with an unmerged change, and a child (new, and the older-release shape bound to the parent's workspace) that runs to `succeeded` in its own task files with a fresh clone at the project base and its own ready repository receipt. The parent's checkout keeps its change. Both cases fail on the previous head with "needs a project / project workspace". ## Risks - **Project-level configuration still applies to children exactly as to top-level tasks.** In a project configured for `shared_workspace`, every task (top-level or child) uses the project's one folder under the existing shared-workspace concurrency policy. A branch that the project itself pins applies to every task in the project. This PR removes only the parent-to-child carry-over of the realized workspace; it does not change those project defaults. - **Standalone follow-ups that an older release bound through `inheritExecutionWorkspaceFromIssueId`** are migrated only when the source is in their recorded lineage (parent or privacy parent). Others keep their explicit share and the existing writer-wait serialization. - **Behaviour change for agents that relied on a shared checkout.** A child or follow-up no longer sees the parent's unpushed or uncommitted changes. Agents must push or merge code they want a subtask to build on. The skill and onboarding text now say this. - **Explicit ancestor workspace ids are silently dropped** instead of rejected, so older clients do not break. The task gets an independent workspace. - **First run of a migrated child starts from the project base.** If an older child had uncommitted work in the parent's workspace, that work stays in the parent's workspace and is not copied into the child's new workspace. - **Admission adds one indexed read** for tasks with a parent and a bound workspace; the locking transaction and ancestor query run only when the bound workspace was created by another issue. ## Model Used - Claude Opus 5.5 (Anthropic), model ID `claude-opus-5-5`, used through Claude Code with tool use (file edits, shell, test execution) and extended reasoning. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [x] All Paperclip CI gates are green - [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
5f1b1bb81d |
fix(recovery): continue from the last save when a retained source cannot be verified (#19867)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - When a native run cannot save its files, a retained-file recovery hold blocks the task until the worker saves them from the retained sandbox. > - The worker verifies the recorded source strictly. This includes the reusable-lease chain of the sandbox. > - A reusable sandbox chain can cross from a parent task into a child task. The check then refuses the source with `allocation_ownership_ambiguous` on every attempt. > - Retry repeats the same refusal. The task stays blocked by three gates: the saved-work wait, the repair flag and the remote writer hold. The only text left tells a person to inspect the run and its environment. > - The source check must stay strict. Recovery must still work without a person. > - This pull request adds an automatic fallback: the task continues from its last committed workspace save, the unverified files are attached to the task but never applied, and the unverified workspace epoch is abandoned so its root is never reused. > - The benefit is that tasks with an unverifiable retained source continue by themselves, with no lost saved work and no admin action. ## Linked Issues or Issue Description Refs #19840 (removes the "inspect the run and its environment" text and keeps Retry; this change removes the need for that Retry). **What happened?** A child task ran in a reusable sandbox that it inherited from its parent task. The run saved a checkpoint, wrote more files, and then failed to save them. Retained-file recovery refused the source with `allocation_ownership_ambiguous`, because the lease history of the sandbox crosses into the parent task. The hold was exhausted after one attempt. Retry made a new generation that failed the same way. The task stayed blocked with `saved_work_recovery_failed`, `workspaceRepairRequired: true`, and the failed run still the dirty writer of the workspace. A second task had a stale exhausted hold for a run whose files a sibling recovery action had already saved. That hold also kept the task from continuing. **Expected behavior** Recovery works without a person. When the source cannot be verified, the task continues from its last saved version. The unverified files are not applied, but they stay available. A stale hold whose files were already saved settles as recovered. **Steps to reproduce** 1. Use a native run with a remote workspace whose reusable sandbox lease history includes a lease of another task. 2. Let the run save a checkpoint, write more files, and fail its final save. 3. Let retained-file recovery run, or press Retry. 4. Before this change, the hold is exhausted with `allocation_ownership_ambiguous` and the task stays blocked. **Paperclip version or commit** Observed on deployed builds that include `630411af53`. Holds saved by the Nov 3–6 releases have this shape. ## What Changed - New `server/src/services/workspace-recovery-fallback.ts`: - `continueFromLastSavedWorkspace` runs when a hold is exhausted by a permanent source refusal: `allocation_ownership_ambiguous`, `source_binding_invalid`, `source_unavailable` or `historical_source_missing` with retry class `service_required` or `use_alternative_strategy`. It applies only to remote workspace sources in the `prepared` state. - Quarantine (secrets omitted under the workspace file policy: `.env`, keys, repository metadata, dependency directories): if the retained root is readable, the worker copies it with the existing read-only capture (no write to the sandbox), converts it to a `.tar`, stores it, and attaches it to the task as `unverified-files-run-<runId>.tar`. A `workspace_recovery_quarantine` operation records it as `verified: false, applied: false`. If the root is not readable, the worker records a closed reason label. - Stop proof: every allocation of the source run must carry a provider stop receipt. A missing receipt is obtained through the same stop-only cleanup as normal recovery (`prepareSandboxStopAndRetain` / `prepareHistoricalSandboxStopVerification`, `retryPendingSandboxTeardown`, `settleStopOnlyCleanup`). A sandbox that another lease holds (`active` / `pending_cleanup`) is never read or stopped from here; the hold waits. - One transaction (lock order: task, coordinator, run, recovery action, workspace) re-checks all facts including the stop receipts, abandons the epoch, resolves the hold with the named outcome `retained_source_unverifiable_continued_from_last_saved`, records an `executionReconciliation` decision with `retainedSourceFallback`, and denies native replacement. A Blocked task belongs to the hold only while the hold's own `workspaceFinalizationBlock` receipt still owns that status version, or, for receipt-less holds saved by older releases, while the block was set by the run's terminal failure (within its finalization window) and no status decision followed. Only then is the task restored to `todo` and a continuation scheduled; a newer block is kept and no continuation starts. The decision carries machine `workspaceRepairEvidence`, which the reconciled admission requires for runs with a restore failure. - Continuation: with no waiting user input, the hold uses `continuationDelivery: "pending"`, so the existing reconciled delivery starts one continuation. Waiting user input continues the task itself (`not_requested`), so there is never a second continuation. - Agent note: the decision's `outcomeEvidence` (part of the continuation's recovery outcomes) says that the run's unsaved changes could not be verified and were not applied, where the attached copy is, that the agent should check what to redo, and that completed external actions must not be repeated. - `settleSiblingRecoveredHold`: an exhausted hold whose run has a verified repair receipt from a different recovery action settles as recovered. It runs for every exhausted hold, including adoptable native sources. A sibling marked recovered whose receipt no longer verifies records a permanent deferral, so discovery does not reselect it. - Deferral: when the fallback cannot settle yet, the hold records `retainedSourceFallbackCheck` (reason, generation, count). Transient reasons (live writer, running operation, missing stop, sandbox in use, pending status effect, lock busy, storage failure) are checked again after 5 minutes, at most 12 times. Permanent reasons (uncertain external action, non-remote source) wait for new input, Retry or a new generation. Storage or database failures after a successful read are retryable and never recorded as "unreadable". - New `server/src/services/workspace-recovery-fallback-discovery.ts`: SQL for pending and due fallbacks. `retainedWorkspaceCandidates` (the production worker's obligations sweep) and `nextWorkspaceRecoveryAt` (its scheduler) include them, so the worker reaches old holds without saved input. - New `server/src/services/remote-workspace-epoch-abandonment.ts` (kept out of `remote-workspaces.ts`, which is near 1,000 lines): authority transition `abandonUnverifiedRemoteWorkspaceEpoch`. It clears the writer and the dirty flag, keeps `snapshotId`, drops `location`, advances the epoch, clears `providerRef` and the migration root, and records the abandoned epoch. The next preparation restores the last save into a new `<workspace>-<epoch>-<lease>` root. - `server/src/services/workspace-recovery.ts`: hooks. An exhausted hold first tries the sibling settlement, then the fallback (under the existing workspace ownership lock). A fresh permanent refusal in an attempt runs the fallback in the same pass, under the same ownership. - `remote-workspace-transport.ts`: export the existing read-only capture. - `packages/shared`: optional `retainedSourceFallback` on `ExecutionReconciliation`. - `doc/execution-semantics.md`: document the fallback. - No new polling or worker: the fallback runs inside the existing retained-workspace recovery pass (`rules/sleep-aware-background-work.mdc`). Safety: - The cross-task lease check (`retiredReusableLeasePredecessors`) is unchanged and stays strict. Unverified files are never applied. - No second executor: the source run, its controller and its finalizer must be stopped. A live writer of the task or workspace, or a running workspace operation, keeps the hold. - No repeated external side effects: pending status effects, non-completed or unknown API receipts, and uncertain tool invocations keep the existing reconciliation path. The continuation keeps the existing API receipt fences. - Never overwrite another run's files: only the exact epoch that the failed run wrote is abandoned, and the old root is never reused. - Bounded and idempotent: a settled hold is never revisited. The quarantine record is reused. A changed generation or revision leaves the hold unchanged. Previous-release story (`rules/upgrade-safe-run-state.mdc`): holds saved by earlier releases carry forward. A hold saved by the Nov 3–6 releases (`systemRecovery` exhausted with `allocation_ownership_ambiguous` / `service_required`, board owner, old admin text) takes the fallback on the next worker pass after deploy (restart sweep, Retry, or a new message). No migration is needed. Refusals that do not qualify keep today's behavior. The only new persisted shapes are additive: `retainedSourceFallback` and `siblingWorkspaceRecovery` on the hold, `remoteWorkspaceAbandonedEpochs` on the workspace, and a `workspace_recovery_quarantine` operation. Earlier releases ignore them. ## Verification - `cd server && npx vitest run src/services/workspace-recovery-fallback.test.ts` — 15 tests: - cross-task hold, saved by the previous release and refused fresh: continues from the checkpoint snapshot, the quarantine `.tar` holds the unverified file, the epoch is abandoned, `getExecutionBlocker` is `null`, the task is `todo`, explicit continuation is not blocked by `saved_work_recovery_failed` or `workspace_repair_required`, one reconciled continuation is delivered, and the next writer restores into a new root while the old root keeps its files; - unreadable root: continues and records the reason; - idempotency: a second pass changes nothing and delivers no second continuation; - waiting user input: no second continuation; - negatives (live writer, pending side effect, uncertain external action, sandbox held by another lease): the hold stays, the authority is unchanged, nothing is attached or delivered, the deferral reason and recheck time are recorded; - an allocation without a stop receipt is stopped (stop-only) before the hold settles; - the production worker (`reconcileRetainedWorkspaces` with `obligationsOnly`) and `nextWorkspaceRecoveryAt` discover and settle an old hold with no saved input; a transient deferral is not revisited until due, then settles; - an old-release hold without a Blocked receipt keeps its status; a current hold with the receipt is restored to `todo`; - stale exhausted hold with a sibling-saved receipt settles as recovered; one whose sibling receipt no longer verifies is recorded and no longer discovered; - a newer Blocked decision is kept and no continuation is delivered, while the file hold settles; - the attached copy omits `.env` and key files. - RED: with `workspace-recovery.ts` and `system-recovery.ts` reverted to master, all tests fail. - Regression suites pass: `workspace-recovery-warm-handoff`, `workspace-recovery-retry-copyback`, `workspace-recovery-preparation-gates`, `workspace-recovery-source`, `task-retry`, `execution-blocker-read` (288 tests), and `workspace-recovery-admission`, `workspace-recovery-preparation*`, `remote-workspaces.integration`, `native-failed-workspace-recovery`, `remote-workspace-transport`, `workspace-restore-repair-route` (683 tests). - `server` typecheck passes. ## Risks - A task now continues without the unverified changes of its failed run. The agent receives a note and the attached copy, so it can redo or re-apply the work. This replaces an indefinite block. - Abandoning an epoch on a workspace shared with a parent task also moves the parent to the last save. This is correct: the unverified changes are not part of the saved lineage. - Clearing the migration root means sessions bound to the old root start fresh. The fallback continuation already forces a fresh session. - The quarantine capture reads a sandbox that can be shared with another task. It writes nothing there, and it verifies the copy against inventories taken before and after. ## Model Used - Provider: Anthropic. Model: Claude Opus 5.5 (`claude-opus-5-5`), with tool use (shell, file edits, read-only browser for API inspection) in Claude Code. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
Dotta
|
eb7c7b083a |
Recover Muse verification after contact, restart and access restoration (#19909)
## Thinking Path > - Paperclip manages AI agents and their work under company access rules. > - Muse readiness uses a saved connection, receiver contact and an authenticated challenge reply. > - A failed challenge write could leave a saved contact with no later recovery trigger. > - A server restart or restored access must recover that contact under current authority. > - This pull request uses the existing delivery coordinator and saved Muse state for recovery. > - It keeps the normal receiver schedule, authorization checks and real deadlines. > - It lets eligible verification recover and lets idle verification stop work. ## Linked Issues or Issue Description **What happened?** The first Muse receiver contact could persist while readiness challenge creation failed. The contact then had no durable retry or startup recovery path. Later membership, lifecycle or feature restoration could also leave pending verification without a notice. **Expected behavior** Retry a temporary failure through the existing worker. Recover pending contact after restart. Recheck the current company, agent, operator, credential and binding generation before issuing a challenge. Stop verification work while it is ineligible or complete. Do not revive revoked credentials. **Steps to reproduce** 1. Use the normal invitation and OAuth registration, PKCE and token flow. 2. Send one receiver contact and let its normal persistence schedule run. 3. Fail the first readiness challenge operation before it commits. 4. Observe whether the existing worker retries without another contact or a manual Verify action. 5. Restart the App with a new database owner. Check that it creates one usable challenge from the saved contact. **Paperclip version or commit** The causal baseline test used the unchanged production implementation at `90148317b2a447794ff29a4acae5648ad11a9770` plus its normal fixture. The accepted positive test used `9d383041028776ad8469ac4af3fe14ba6c369425`. The initial published head was `fe518c9b72504daf6374225987f24d45bc7bcf75`. The lease/auth correction was locally qualified at `cfd515a207416097fe8b586277d70c4fd4a679fa`. The current candidate is `6efe722e7eb1e22acd466125c7b4a627611d0273`; its added Cloud actor fixture correction passed the full normal owning program and all 35 tests. Its current normal build, lifecycle owner controls, healthy-wait and restoration measurements are described below. Earlier results below are bound to their stated revisions. Related public changes: Refs #19819 for original-principal authentication. Refs #19750 for the transaction-owned external feature read. These are related foundations, not duplicate fixes. The current Muse title and head-branch searches found no duplicate contact-verification recovery PR. Related maintenance work is #17000 and terminal assignment settlement is #17793. ## What Changed - Add a typed, exact-binding verification notice through the existing Muse delivery queue. - Register the first-contact notice before its durable write in the same transaction. - Recover pending contact from saved state at startup. Retry temporary failures through the existing scheduler. - Recheck current feature, lifecycle, credential and access state after commit. Park unavailable authority without recurring verification scans. - Add narrowly scoped notices to the existing access, lifecycle, configuration, company, budget and settings write owners. - Save a healthy lifecycle verification wait in the existing operation JSONB. Keep generic unresolved-work and error retries unchanged. - Commit authenticated confirmation and its lifecycle recheck in one transaction. Retain a raced recheck under the active lease, and recover saved waits once at startup. - Retain a committed lifecycle recheck while an expired attempt still owns the row, so its late wait cannot park accepted confirmation. Replacement claims keep their existing owner fence and fresh fact read. - Keep the receiver gap, persistence and per-segment contact limits unchanged. - Add real App OAuth, failure, retry, restart, challenge and quiet evidence. Add two normal-route restoration tests. - Document recovery, reconnect requirements and the existing connected-receiver sleep restriction. ## Verification - Causal baseline: the real invitation and OAuth flow saved one contact. The first challenge operation failed. The normal test failed because it observed one verification attempt instead of a retry. No setup or cleanup failure was counted as this regression. - At `9d383041`: normal own Shared/SDK, DB with migration 0441 assets, Runner TypeScript core and surfaces, module boundaries, and all seven normal App programs passed. One new redundant role comparison first failed typecheck; that diagnostic remains recorded and the corrected full App gate passed. - At `9d383041`: all 17 maintenance component tests passed with zero skips. All original deadline, rollback, drain, terminal settlement and startup controls remain. - At `9d383041`: the one normal App case passed. Real OAuth authorized one receiver signal. A retry occurred about 5.01 seconds after the first failure. A new Db/App owner created one challenge, advanced revision 2 to 3, and accepted authenticated confirmation and mailbox ACK. Query instrumentation had an active positive control. All measured query, callback, session and verifier counters stayed equal across the unchanged scheduled 15-second quiet window. Token revocation and old-token/signal rejection passed. - Known App, worker, HTTP and database owners closed without cleanup errors. The tests used bounded local PostgreSQL and HTTP fixtures. They did not call a live Muse model. - Current integrated semantics: the full normal App driver completed all programs. Only a new test callback had two undefined-value diagnostics. An explicit defined-value guard corrected that test. The full 416-root owning program then passed with empty diagnostics. A separate collector incorrectly required an explicit child `noEmit` field; the normal inherited setting was verified without repeating the compiler. The original failures remain recorded. - The first restoration run failed both cases. Permission parking recorded 120 main-pool queries without more verifier calls. The lifecycle counter recorded two denials without their exact scope. These observations remain unexplained at the individual-query/call level. Source separately shows a lifecycle readiness retry every two seconds while the real external reply is absent. - At `fe518c9b`: normal Shared/SDK and DB producers, module policy, all seven full App programs, and all 33 pure diagnostic controls passed. The lifecycle owner file passed all 19 cases, including both confirmation/park orders, rollback, duplicate notices, startup recovery and authority controls. - At `fe518c9b`: the normal OAuth failure/retry/new-Db reconstruction case passed. Its new healthy-wait window measured 15,002 ms with unchanged total activity before genuine confirmation; the existing ready quiet and revocation checks also passed. - At `fe518c9b`: both normal restoration tests passed in Vitest. Ordinary viewer-to-operator restoration and Pause/Resume recovered the same binding, grant and single contact; each committed one challenge and advanced revision 2 to 3. All six parked, restored and revoked/disabled quiet windows measured at least 15 seconds with active positive read controls and unchanged total queries, callbacks, sessions and verifier calls. Both cleanups passed. - The restoration wrapper still recorded failure: it required lifetime lifecycle denial count zero, whereas the measured count was one before the parked quiet window. The reviewed fixture explicitly permits a current-authority denial from an earlier observation and requires no committed challenge before restoration. Precise denial timing relative to Pause was not serialized. The original wrapper failure is retained; no rerun or retrospective explanation of the earlier failures is claimed. - Review exposed an expired-lease race. The added real-store baseline test failed after the late defer restored `waitingForSignal`; no setup or timeout was counted as the regression. On `cfd515a`, the whole lifecycle owner file passed all 20 tests, including that case. The full normal production declaration program (1067 roots), verification program (662), and auth owning program (405) passed; all 1942 declarations remained byte-identical. - The original remote auth-session shard failed three tests because their partial query adapters lacked the new binding query's join/order/limit chain. A fixture-only correction preserved the real notifier and existing authorization assertions. All 12 auth-session tests passed locally. - Complete remote inspection also found 33 failures in the 35-test Cloud actor fixture with the same missing `innerJoin` method and downstream incomplete-auth assertions. Its original full log and exact synthetic Source comparison are retained. At `6efe722e`, the three-line fixture adapter repair passed the complete normal 662-root test program and the whole 35-test file with zero failures or skips. All existing lock, authorization, settings, debounce and unfinished-auth assertions remain. - The earlier healthy/restoration quiet measurements remain bound to `fe518c9b`. Those steady 15-second windows did not exercise a still-owned attempt passing its 120-second lease. The targeted expired-owner regression qualifies that new boundary; it is not a fresh measurement of the earlier windows. - The initial `fe518c9b` remote gates failed the auth fixtures and reported a Greptile 4/5 P1. The aggregate failure reported the underlying shard. Those actual causes were fixed and qualified above. - Current `6efe722e` remote gates completed with **57 successful checks and three explicit skips**. The full latest edited Greptile summary is **5/5**, explicitly reviews this exact head, and has no new finding. The original P1 thread is resolved. All general, inline and nested review pages were read. - One current-head CI attempt timed out after 30 seconds in the Runner evidence/privacy file, with 6,864 passes and one failure. That file and its transitive inputs are byte-identical to the CI base. A separate local diagnostic preserved every assertion and the original timeout, ran the full normal Runner test project and the entire 41-case file, and passed all 41 with zero skips. Its private bounded timing records measured the target body at 74 ms. The diagnostic was then removed, restoring the exact original test. This is local recovery evidence; the original CI stalled await was not recorded and its cause remains **unproven**. - One retry of only the failed jobs on the unchanged head passed the Runner support suite and dependent aggregate verification. The complete second attempt and original failure logs are preserved. No test, timeout, privacy guard or product code was weakened. This result does not turn the original timeout into an explained failure or a diagnosed flake. - Full recursive typecheck, full `test:run` and a full release build were not repeated locally. The normal full App programs, focused producers and tests above passed, and the required current-head remote gates passed. Final review is against the exact unchanged head; local quiet remains separate from deployed suspension. Reviewer commands use the normal project entry points: `node server/scripts/typecheck-layers.mjs`, the whole `server/src/services/muse-maintenance.test.ts` file, and both normal App PostgreSQL files. PostgreSQL fixtures require explicit owned fixture/admin URLs and normal migrations before imports. Run the files serially with one worker. The private controller retains exact Source, output, worker and owner-closure receipts. ## Risks - Missing or early producer notices could lose a restoration trigger. Notices stay in the transaction that commits the write. The worker checks current authority after commit. - A stale notice must not grant access. Binding generation, current membership, company, agent and credential checks remain authoritative. Revoked credentials require reconnecting. - The change adds no table, migration, alternate task queue or periodic verification scan. It adds two optional fields to existing lifecycle operation JSONB. Generic pending/error deadlines, setup holds and manual repair remain obligations. - Confirmation can race with an active verification attempt. The park compare-and-set must retain the durable recheck. Startup recovery uses a fixed cutoff and keyset pages; it must not become another recurring scan. Wrong company, stale version, pause and revoked credentials must stay fenced. - The connected receiver still has its documented tenant-sleep restriction. Worker quiet is not proof of deployed suspension or external wake. - The normal challenge reply is sent by an authenticated protocol fixture. Live Muse background replies, cost/usage, full Muse lifecycle and full hosted Dot behavior remain unqualified. Partial Dot profile ACK and unsigned-device cutover stay withheld. - Restoration coverage for the additional company, budget, configuration and feature producers remains a separate qualification obligation. This pull request does not mark the whole hosted area complete. ## Model Used OpenAI `gpt-6.1-sol` authored the implementation and tests with local code and tool use. A GPT-6-based Codex coordinator reviewed the changes and owns integration. Its exact underlying model ID is not exposed. Context window sizes are unavailable. No context size is inferred. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [x] All Paperclip CI gates are green - [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge |
||
|
DottaandClaude Opus 5.5
|
cbf811cd3b |
fix(recovery): prove a replaced deployment's copyback owner gone automatically (#19913)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - Native runs save their workspace after the result is accepted, under
a copyback owner receipt
> - A deploy can restart the server while that save is in progress
> - The new server runs in a new container, so it cannot see the old
process and fails closed
> - The run then waits on a board action that asks a person to verify
the previous controller
> - Recoveries must continue without a human step, and copy must not ask
anyone to inspect a server
> - This pull request adds a deployment-level rule to the shared
owner-liveness primitive and uses it in workspace finalization
> - The benefit is that a save interrupted by a deploy resumes on its
own, with no second writer
## Linked Issues or Issue Description
**What happened?**
A deploy restarted the server while three runs were saving their
accepted results. Each run kept a copyback owner receipt that names the
old Railway deployment on another container. The restarted server
reported `native_workspace_finalization_owner_unverified`, owned by the
board, with the next action "Verify the previous controller and its
workspace-copyback processes have stopped...". The cloud harness stop
receipt did not arrive, so the runs stayed in "Recovery needed" for
about an hour, until a person confirmed the stop by hand.
**Expected behavior**
The restarted server proves that the previous controller is gone,
resumes the save, and settles the run with no human step. No user-facing
text asks a person to verify a server or controller.
**Steps to reproduce**
1. Start a native run on a Railway service with a volume.
2. Let the result be accepted and the workspace save start.
3. Deploy, so that the server restarts during the save.
4. See that the run waits for a board confirmation.
**Paperclip version or commit**
Observed on
|
||
|
DottaandClaude Opus 5.5
|
99529f9d71 |
feat(observability): local speed budgets and a run-log budget check (#19863)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - When someone messages an agent, the reply must start as fast as
possible, and restart-safety work must not slow it
> - The durable ownership plan makes this invariant 8 ("fast turns") and
asks Phase 1 to turn the Phase 0 baseline into p50/p95 budgets per
adapter
> - Without budgets, a regression in time to first token, agent boot or
restart recovery shows up only when someone reruns the baseline by hand
> - This pull request records the budgets and adds a per-run check that
writes a run-log row when a run exceeds its p95 budget
> - The benefit is that every later phase can show "no regression
against the budgets" from the run log, without new waits on the turn
path
## Linked Issues or Issue Description
Refs #19831 (plan, "Performance constraint", rule 4, and invariant 8).
Refs #19846 (the Phase 0 baseline these budgets come from). Stacked on
#19861.
**Problem**
The plan requires p50/p95 budgets per adapter for TTFT, agent boot and
restart recovery latency, and a check against them. Neither exists.
**Expected behavior**
Budgets are recorded in the plan and in code. A run that exceeds a p95
budget gets one run-log row that names the metric, the measurement and
the budget.
**Impact**
Phase 2–5 exit criteria can cite budget breaches from chaos waves and
staging. The check adds no wait before `turn.start`.
## What Changed
- New `server/src/services/native-runtime/native-speed-budgets.ts`: the
local budget table (codex, claude, opencode) for TTFT and agent boot
(cold and warm) and restart recovery latency. Each budget is 1.2× the
baseline, rounded up to 0.5 s. `evaluateSpeedBudget` judges one run
against p95. An unknown start kind uses the looser budget. Remote runs
and adapters without a baseline are not judged.
- `native-run-trace.ts`: the trace takes an optional `budget` subject.
It checks `task.time_to_first_token`, agent boot (end of
`heartbeat.queue` to the end of the first `runner.turn.submit` or
`runner.session.resume`) and restart recovery once each. It writes
`run.performance.budget_exceeded` through `defer`, so the write is
detached from the event path. `finish` now drains diagnostics that defer
more work.
- `native-session-executor.ts`: passes the adapter and environment to
the trace, labels cold or warm at the existing `provider_session` reuse
receipts, and checks restart recovery at a recovered run's first
provider event, measured from the startup recovery `ready` time.
- Docs: budget tables in
`doc/plans/2026-11-07-durable-ownership-and-reconciliation.md` ("Speed
budgets (Phase 1)"), and the new event in `doc/run-log-events.md`.
## Verification
- `npx vitest run
src/services/native-runtime/native-speed-budgets.test.ts
src/services/native-runtime/native-run-trace.test.ts` (in `server/`): 34
tests pass. They cover:
- every budget has p50 ≤ p95;
- the provider-to-adapter mapping;
- p95 judgement, and the looser budget when the start kind is unknown;
- one breach for a slow boot, and none for a second submit or a fast
TTFT;
- a deferred TTFT breach is written before the trace closes;
- restart recovery is checked once;
- remote runs and runs with no budget subject write nothing.
- `npx tsc --noEmit -p server`: no errors.
## Risks
- Low risk. This is a run-log change only (AGENTS.md §5.7, no extra
review). The OTLP and Telemetry paths are unchanged.
- The baseline samples are small (2–12 per cell) on a shared host.
Budgets carry 20% headroom, and the codex restart budget (8.0 s p95) is
tight. Expect some breach rows on a loaded machine. They are diagnostics
and change no behavior.
- Restart-recovery budgets use run 4 of the baseline, as merged in
#19846.
- Restart recovery is measured from the startup recovery `ready` time to
the first provider event, which matches the baseline definition.
- Previous-release story: no persisted run state changes. The new event
is additive.
## Model Used
- Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, in Claude Code with
tool use and code execution.
## Checklist
- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have searched GitHub for duplicate or related PRs and linked
them above
- [x] I have either (a) linked existing issues with `Fixes: #` / `Closes
#` / `Refs #` OR (b) described the issue in-PR following the relevant
issue template
- [x] I have not referenced internal/instance-local Paperclip issues or
links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip`
URLs)
- [x] My branch name describes the change (e.g. `docs/...`, `fix/...`)
and contains no internal Paperclip ticket id or instance-derived details
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [ ] All Paperclip CI gates are green
- [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [x] I will address all Greptile and reviewer comments before
requesting merge
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
||
|
DottaandClaude Opus 5.5
|
d5dd09c966 |
docs(plans): design turn-scoped ACPX MCP authority; correct the Claude baseline (#19912)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - Follow-up messages must start fast. #19888 made Codex and OpenCode follow-ups 2.6–5× faster by keeping their runners warm. > - Claude (ACPX) did not gain, because ACPX relaunches the Claude process and reloads its session on every follow-up (3.2–4.2 s). > - That relaunch is a safety boundary: the ACPX MCP endpoint lives as long as the process and cannot attribute a tool call to a turn, so reaping the process is the proof against late calls. > - Removing it needs a new authority model, which needs a reviewed design before any code. > - This pull request adds that design and corrects the Claude baseline. ## Linked Issues or Issue Description Refs #19888 (warm default), #19846 (TTFT baseline, finding 4). **Docs issue** The TTFT docs did not explain why Claude follow-ups relaunch, what removing that would require, or the corrected Claude numbers. **What is wrong or missing** `doc/plans/2026-11-07-warm-follow-up-ttft.md` reported a Claude follow-up at 10.5 s, measured under heavy host load. A quieter re-measurement gives 6.3 s per-turn against 7.1 s warm. There was no design for removing the per-turn relaunch. **Proposed fix** A design document for turn-scoped ACPX MCP authority, and a corrected-baseline section in the warm follow-up doc. ## What Changed - New `doc/plans/2026-11-07-turn-scoped-acpx-mcp-authority.md`, which covers: - the measured cost breakdown; - how the current boundary works (tool bridge bearer, call-ID tombstones, `restart_idle_provider`, `run.attach`); - a threat model (late tool calls from a settled turn, replay, credential replay, cross-turn and cross-run confusion, restart replay); - three candidate designs with their safety arguments: turn-scoped credentials in a live session, a proof that no calls are in flight, and an in-process session reload with per-generation credentials, which is recommended for a spike; - restart and chaos implications, the expected TTFT gain, and next steps. - `doc/plans/2026-11-07-warm-follow-up-ttft.md`: corrected Claude baseline. Per_turn stays the default; the earlier 10.5 s was host noise. Finding 4 now points to the design and drops the idle-retirement idea, which would only bring warm level with per-turn. - Docs only. No code changes until Dotta has reviewed the design and a security review is done. ## Verification - The Claude numbers come from `ttft-baseline.ts --adapters claude --samples 6 --phases coldwarm --lifecycle <per_turn|warm>` on master `fc6e1b0346` (load average about 9). - The step breakdown comes from temporary timing in runnerd's ACPX backend. It was not committed. - `node scripts/check-maps.mjs` passes. ## Risks - Docs only. The design proposes no behavior change until it has been reviewed. > For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and discuss it in `#dev` before opening the PR. Feature PRs that overlap with planned core work may need to be redirected — check the roadmap first. See `CONTRIBUTING.md`. ## Model Used - Anthropic Claude Opus 5.5 (`claude-opus-5-5`) in Claude Code, with tool use (shell, file edit, test execution). ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
177f642da7 |
fix(recovery): continue native runs that failed before their first claim (#19911)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - Native runs go through a coordinator claim before any provider work
starts
> - A run can fail before that claim, for example while it restores the
previous harness backup
> - The failure path then records a no-replay hold, as if the run had
acted
> - Workspace recovery saves the files, but the task stays blocked and
asks a person to send a message
> - Recoveries must continue without a human step when no side effect is
possible
> - This pull request lets the existing safe-replacement worker continue
such runs once, after their files are saved
> - The benefit is that these tasks resume automatically, and runs that
may have acted keep today's reconciliation
## Linked Issues or Issue Description
**What happened?**
Native runs that failed before their first coordinator claim were never
continued automatically. They had no claim, no provider process and no
provider or tool events. Their result recorded `executionRecovery: {
kind: "bootstrap", providerWorkStarted: false }`. The failure path still
moved the coordinator to `terminal_failure` with
`native_continuation_requires_reconciliation` and recorded a resolved
`preserve_without_replay_v1` hold (replay blocked, board owner).
Workspace recovery then saved the files, but the next action stayed
"Saved work recovered. Send a message to continue."
`nativeReplacementPending` accepted only a few failure codes, and the
interrupted-preparation path covers only legacy runs.
**Expected behavior**
When a native run provably failed before its first claim, Paperclip
continues it once, automatically, after its files are saved (or when it
had no workspace to save). Runs with a claim, a process, a provider or
tool event, or recorded provider work keep the existing reconciliation.
**Steps to reproduce**
1. Start a native run with a remote workspace.
2. Make it fail before `native.coordinator.claim` (for example, abort
the harness-backup restore).
3. Let workspace recovery save the files.
4. See that the task stays blocked and asks for a message.
**Paperclip version or commit**
Observed on
|
||
|
DottaandClaude Opus 5.5
|
fc6e1b0346 |
perf(native): keep local Codex and OpenCode runners warm by default (#19888)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - When someone messages an agent, the turn must start fast. Time to first token (TTFT) is a first-class metric. > - The Phase 0 baseline (#19846) found that every local turn launches a new runner and loads the provider session again, because `per_turn` is the default lifecycle. That made a follow-up message slower than a new task. > - Warm runners keep the runner and provider session alive between runs. #19874 fixed the defect that broke warm OpenCode follow-ups, and #19879 bounds an idle warm runner whose controller dies. > - This pull request makes warm the default for local Codex and OpenCode agents, and migrates existing agents that only carried the old creation default. > - The benefit: follow-up TTFT p50 drops from 12.0 s to 4.7 s (Codex) and from 21.7 s to 4.2 s (OpenCode), with no new restart-chaos failures. ## Linked Issues or Issue Description Refs #19846 (TTFT baseline), #19852 (startup sub-spans), #19874 (warm completion contract), #19879 (warm reconnect grace). **What existing behavior does this improve?** The default runner lifecycle for native agents, and how a follow-up message starts. **Current behavior** Agent creation writes `lifecycleMode: "per_turn"` into every native agent. Every follow-up launches a new runner and restores the provider session: 4.2 s (Codex) and 16.3 s (OpenCode) at p50 before the turn starts. **Proposed behavior** New native agents store no lifecycle. On a local target, an unset lifecycle resolves to `warm` with the default 5-minute idle window for Codex and OpenCode. Other harnesses, remote targets and managed providers without warm support stay `per_turn`. An explicit choice always wins. Migration 0443 clears the stored creation default from existing agents, except where an edit set "Turn by turn" explicitly. **Reason and benefit** Measured locally (`doc/plans/2026-11-07-warm-follow-up-ttft.md`). Warm follow-ups are 2.6–5× faster for Codex and OpenCode, and restart recovery latency is equal or lower. ## What Changed - `native-lifecycle.ts`: an unset `agentMode` resolves to warm (5-minute idle window) for local `codex_app_server` and `opencode_server` agents. `runtime-selection.ts` passes `localTarget`. Explicit `per_turn` or `warm` wins. An admitted session keeps its persisted policy. ACPX Claude stays `per_turn`: it relaunches its provider process on every follow-up by design, so warm did not shorten its follow-up (10.7 s warm vs 10.5 s per_turn). - Creation and configuration writes no longer store `lifecycleMode: "per_turn"` for Codex, Claude or OpenCode. This covers `normalizePaperclipRunnerAdapterConfig` (runs on every create and update), `paperclipRunnerTransitionConfig` (switching to the native runner), `agent-runner-selection.ts`, `codex-local` `build-config.ts`, and the simplified new-agent setup (`paperclipRunnerCreationLifecycle` in `@paperclipai/shared`). Grok, Pi and Cursor keep their explicit value. - Agent form: the Runner lifecycle field shows the unset state as "Automatic". "Turn by turn" and "Warm session" remain. - Migration `0443_native_warm_follow_up_default.sql`: clears `lifecycleMode = 'per_turn'` from native Codex, OpenCode and Claude agents unless `agent_config_revisions` shows an edit that set it to `per_turn`. Each migrated agent gets an `agent.updated` activity entry with the old value (`actor_id = migration:native-warm-follow-up-default`). Creation records no revision, so a "Turn by turn" pick made at creation cannot be told apart from the old default. The product owner approved migrating those agents. Claude agents follow the default, which is `per_turn` today. - Restart chaos harness and TTFT driver: `--lifecycle per_turn|warm|default` (saved in the wave plan). `default` removes the catalog's pinned lifecycle, so a wave exercises the real default path. - Docs: `doc/DEVELOPING.md` "Runner lifecycle default", and the results in `doc/plans/2026-11-07-warm-follow-up-ttft.md`, which also documents baseline finding 4 (Claude's resumed turn-start gap is ACPX's per-turn process generation, not the provider). ## Verification - `server`: `native-lifecycle.test.ts` (new default cases: local Codex and OpenCode warm, Claude, remote, unknown target, unmeasured harness, managed Claude, admitted per-turn session, explicit choices, non-native adapter), `agent-runner-selection.test.ts`, `runtime-selection.test.ts` (335 pass). - `packages/db`: `native-warm-follow-up-default-migration.test.ts` on embedded Postgres. Default agents are cleared, the explicit edit is kept, other harnesses and legacy adapters are untouched, audit rows are written once, and a rerun is a no-op. Without the explicit-edit filter the test fails. `check:migrations` passes. - `codex-local` `build-config.test.ts` (65), `ui` `config-fields.test.tsx` (40), restart-chaos unit tests (134), `tsc --noEmit` for `server`. - `adapter-utils` `paperclip-runner-lifecycle-default.test.ts`: configuration writes and adapter conversion store no lifecycle for Codex, OpenCode and Claude, keep `per_turn` for the others, and preserve explicit choices. It fails without the fix. - End-to-end default check on this branch (`ttft-baseline.ts --lifecycle default`, codex and opencode, 2 samples): the agents read back with no stored lifecycle, and every follow-up reused the runner and provider session (no `runner.process.launch`). Follow-up TTFT was 3.6–4.8 s. This check found the per-write `per_turn` injection fixed here. - Restart chaos (codex and opencode, A1/B1/B5, K1/K2/K6, seed 4242), 8/8 PASS each with no open signatures: forced warm, forced per_turn, and `--lifecycle default` on this branch (agents created without a lifecycle). - `pnpm verify` selects broader CI (harness config is a global input), so required CI supplies the broad coverage. ## Risks - Behavior change: local Codex and OpenCode agents keep a runner and provider process for up to 5 minutes after a run. Instance sleep waits for that idle window. The process is reaped by the idle timer, by graceful shutdown, or by the #19879 reconnect grace if its controller dies. - Migration: agents whose "Turn by turn" was chosen at creation become Automatic. The activity entry records the old value; choosing "Turn by turn" restores it. - Previous-release story (`rules/upgrade-safe-run-state.mdc`): a task session keeps the lifecycle policy it was admitted with, so sessions saved before this change stay `per_turn`. New sessions get the default. No run, checkpoint or launch state changes shape. > For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and discuss it in `#dev` before opening the PR. Feature PRs that overlap with planned core work may need to be redirected — check the roadmap first. See `CONTRIBUTING.md`. ## Model Used - Anthropic Claude Opus 5.5 (`claude-opus-5-5`) in Claude Code, with tool use (shell, file edit, test execution). ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used (with version and capability details) - [x] I have checked ROADMAP.md and confirmed this PR does not duplicate planned core work - [x] I have searched GitHub for duplicate or related PRs and linked them above - [x] I have either (a) linked existing issues with `Fixes: #` / `Closes #` / `Refs #` OR (b) described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip` URLs) - [x] My branch name describes the change (e.g. `docs/...`, `fix/...`) and contains no internal Paperclip ticket id or instance-derived details - [x] I have run tests locally and they pass - [x] I have added or updated tests where applicable - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |