mirror of
https://github.com/paperclipai/paperclip.git
synced 2026-10-29 12:59:13 +01:00
cff331f6ca0a652b5e41641c68dfa5fa584c22ae
10
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
|
3cd3af5d75 |
Trace native preparation and OpenCode execution phases (#17918)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - Each native run prepares its workspace and skills, then starts its provider. > - A long preparation span does not show whether inventory refresh, a cache lock, repository reads, or file checks caused the delay. > - One hosted run spent 81 seconds in this phase. The same deployed service and pinned files took 2.55 seconds in a separate diagnostic process. > - Another hosted turn spent 69 seconds before its first output. A matched public-workspace local model replay took 7.64 seconds. > - We need skill preparation and provider-internal measurements before choosing another fix. > - This pull request adds skill spans, a real-service benchmark, and opt-in filtered provider timing. > - The benefit is a faster and more precise diagnosis loop. ## Linked Issues or Issue Description Refs #17620. That PR owns the batched read optimization. This PR adds diagnostics and does not change its read policy. Refs #17907. That PR protects long cache lock transactions from the database idle timeout. **What happened?** Native skill configuration can take much longer than matched standalone resolution. Provider execution also has long gaps with no internal timings. Existing spans cannot separate inventory refresh, cache acquisition, repository reads, context loading, Git snapshots, or model streaming. **Expected behavior** An operator-configured trace shows these phases under the existing run context. Without an OTLP endpoint, the new spans stay off. A local benchmark can reproduce the service path with real PostgreSQL, files, an agent actor, and a responsible user. **Steps to reproduce** 1. Run `pnpm --filter @paperclipai/server exec tsx scripts/benchmark-native-skill-preparation.ts --actor=agent --staging-shape=true`. 2. Inspect cold and warm phase timings, SQL counts, and file verification results. 3. Set the existing operator OTLP configuration and add `--export-otlp=true` to inspect the same hierarchy in the collector. ## What Changed - Add bounded skill inventory, refresh wait, cache acquisition, row, version, repository, file check, and consumer spans. - Keep the existing endpoint gate, attribute allowlist, cache locks, authorization, file validation, and durable pin boundary. - Add a standalone benchmark with a full repository fixture, immutable snapshot input, calibrated local transport delay, and optional OTLP export. - Verify every delivered file and executable bit outside the timed section. Create its fixture agent through the lifecycle API. - Add an opt-in authenticated loopback sink for OpenCode. Discard raw attributes, logs and content. Retain only closed phase names and timestamps. - Carry one bounded diagnostic batch through the real proxy, Rust normalization, durable journal and server trace projection. Group provider phases under their execution parent. - Freeze the Date clock in an existing delegated Slack rate-limit test. This fixes a CI failure when consecutive requests cross a rate-window boundary. It does not change rate-limit behavior. - Document phase boundaries and the difference between SDK startup and confirmed ingestion. ## Verification - Skill service, runtime selection, concurrency, cache and timing suites: 122 tests passed with real PostgreSQL and migrations. Enabled-span integration tests also passed. They verify parent relationships and a real advisory-lock timeout. - A captured 208-file repository produced 209 verified runtime files. The benchmark uses real agent and responsible-user authorization. The lifecycle fixture correction passed with the staging-shaped inventory. - Affected OpenCode driver, collector and transport tests: 72 passed. The tests cover authentication, content removal, size/count bounds, disabled behavior and the rebuilt native Runner boundary. One Rust projection test and 18 server trace tests also passed. A held run-log sink proves that a full 128-span diagnostic batch adds no duplicate database writes before terminal consumption. - The two CI-failing delegated Slack rate-limit cases passed against real PostgreSQL with only the Date clock frozen. - Runner TypeScript and server TypeScript checks passed. Rust formatting and module-boundary checks passed. - Real Honeycomb readback confirmed the 14-span skill waterfall and a 40-span provider waterfall. Provider phases have the correct execution parent. The provider export uses synthetic model replies and is instrumentation proof only. - Earlier full local checks failed on missing package-local executables, then on the default Node heap and a missing generated service worker. The dependency links were repaired. The server check passed with a larger heap. The full local test attempt could not start embedded PostgreSQL on this host and was stopped after retaining its failures; real database coverage uses an isolated database on an existing local server. These attempts do not count as a full local pass. - CI on the prior revision passed typecheck, build, Runner and all E2E shards. It failed the benchmark lifecycle boundary and one existing rate-window test. Those fixes passed CI. A later Runner assertion exposed the Copilot source identity pin on the shared transport file. The diagnostic flag now uses the existing configured-environment projection, preserving that file byte-for-byte. All 66 selected installation and native transport tests passed. New-head CI and review remain required. - The under-15-second hosted response target remains unmet. This PR adds diagnostics and does not claim a hosted latency fix. ## Risks - Skill spans and Honeycomb export retain the operator endpoint gate. OpenCode phase collection requires a separate explicit diagnostic flag. Without an endpoint, its closed measurements stay in the instance run log. - Raw OpenCode OTLP can contain prompts and authorization attributes. The new sink rejects unauthenticated requests and discards all attributes before any measurement reaches PRP or the operator collector. Request, event and per-phase limits bound memory and journal growth. - Collection does not wait for a provider flush. Late or oversized exports can be absent; these measurements are not a complete profiling guarantee. - The benchmark's inventory contents and transport delay are explicit fixtures. They do not reproduce server contention or prove hosted response time. - The one durable diagnostic batch remains the local evidence. Its child span export does not add per-phase run-log rows. - No database migration, authorization policy, model selection or inference transport changes. The Rust Runner and OpenCode proxy must be shipped together for the new opt-in diagnostic. ## Model Used OpenAI GPT-6 through Codex, with repository inspection, shell execution, browser tools, and Honeycomb queries. The session does not expose a verifiable backend deployment ID or 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 - [ ] 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> Co-authored-by: Paperclip <paperclip@paperclip.ing> |
||
|
Nicky LeachandPaperclip
|
a3e8fa568a |
Reduce repeated work during plugin Runner admission (#17377)
## Thinking Path > - Paperclip manages agents and their execution authority. > - A plugin can host Runner tasks on a persistent execution host. > - Admission checks must observe current lifecycle state and credentials. > - The current runtime reader also resolves files, skills, and connections for each check. > - Some providers also repeat the same project authentication check during submit. > - This pull request adds a compact fresh reader and an explicit submit-time validation contract. > - Providers can remove repeated work while Core keeps lease and scope checks. ## Linked Issues or Issue Description **What existing behavior does this improve?** Plugin Runner startup before task submission. **Current behavior** One measured startup made 30 full runtime reads during bootstrap. These reads took 5.759 seconds. Each guard can resolve instructions, skills, and connection assignments even when it only needs current identity and lifecycle state. Providers that already enforce project authentication in submit must also perform a separate Core preflight. **Proposed behavior** Use a lifecycle-owned runtime identity reader for guards. Keep the full runtime reader for synchronization inputs. A trusted provider can declare `projectAuthCheckAtSubmit: true` in its prepared result. Core pins the exact configuration, provider, and project set on the lease. The provider must reject authentication conflicts before credential installation or task start. Providers that omit the flag retain the existing preflight. **Reason and benefit** Estimated savings with a paired consumer: 4–5 seconds from compact runtime checks, plus roughly 0.5–0.9 seconds from removing the duplicate authentication operation. These estimates overlap and are not a measured deployed result. Refs #17367 for the separate run-log HEAD optimization; this PR does not duplicate it. **Breaking changes** No migration. The SDK adds a host method and an optional prepared-result field. Consumers that use them require this Core host version. Older providers retain their existing flow. ## What Changed - Add `agents.getRuntimeIdentity` across SDK, RPC, host service, and test fixtures. - Reuse the existing company, plugin ownership, readiness, and capability checks. Read lifecycle state and protected instance secrets fresh on every call. - Skip instruction, organization, skill, and connection hydration in the compact reader. - Allow explicit submit-time authentication enforcement, with prepared scope pinned and checked again at dispatch. - Emit fixed preparation, project-auth, submit, and connection timing phases. ## Verification - 72 targeted SDK capability, plugin task-owner, and environment-task mediator tests passed. - 33 real embedded-Postgres lifecycle and secret tests passed, including pause, content revision, company isolation, plugin disablement, and secret rotation. - SDK and server TypeScript checks passed. Local Runner/eval TypeScript outputs were built to satisfy server type resolution. - Local Postgres required missing library aliases in ignored dependencies; no product source was changed for that repair. - Full repository suites are delegated to GitHub CI. No live rollout or speedup claim. ## Risks The submit-time flag delegates enforcement to a trusted provider. It is not proof that prepare already checked authentication. The provider must check before VM credential installation or task admission. Core still checks exact prepared scope, active lease, run identity, credential type, and dispatch fencing. Existing providers keep the separate check. Roll back a paired consumer before rolling back this host. ## Model Used OpenAI GPT-6 (Codex), with code execution and repository tools. Exact deployment build and context-window size are not exposed. ## 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 described the issue in-PR following the relevant issue template - [x] I have not referenced internal/instance-local Paperclip issues or links - [x] My branch name describes the change and contains no internal ticket id - [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> |
||
|
DottaandPaperclip
|
6c9b87d092 |
trace(server): expose repository admission waits (#17040)
## Thinking Path > - Paperclip manages work performed by AI agents. > - The controller prepares a task workspace before it starts a provider turn. > - Repository admission also runs when a checkout already exists. > - One observed reuse took 16.877 seconds. Its child spans explained only 3.556 seconds. > - The missing detail covered repeated authority checks, catalog work, and publication waits. > - This change exposes those operations under the existing execution attempt. > - Operators can expand the slow phase before choosing a performance fix. ## Linked Issues or Issue Description Refs #16963, #16330, and #16515. The provider-routing work in #17019 has a different scope. **What existing behavior does this improve?** Repository-admission traces for new and reused checkouts. **Subsystem affected** server/ orchestration and operator-configured OpenTelemetry tracing. **Current behavior** The admission span can contain a long interval without child spans. Existing project checks, HTTP requests, and checkout lookup do not explain current-task checks, database awaits, credential refresh, or publication locks. **Proposed behavior** Expose nested spans for current binding and task access, individual identity reads, catalog construction, credentials, grant transactions, filesystem validation, Git inspection, and directory-lock acquisition, work, and release. **Reason and benefit** The trace can distinguish repeated checks, remote calls, database awaits, and directory contention. A database-await span includes driver, network, pool, and server time. It is not isolated SQL execution time. **Breaking changes** None. No API, schema, permission, retry, lock, or repository behavior changes. ## What Changed - Add fixed admission phases beneath the existing execution attempt. Keep the existing project, checkout, and HTTP detail. - Trace current-task authorization stages only under workspace access. Unrelated decisions cannot consume that detail budget before admission. - Trace GitHub catalog reads, credentials, provider metadata, grant row locking and writes, transaction settlement, and webhook work. - Separate directory-lock acquisition, the protected operation, and release without changing the adapter lock contract. - Document the hierarchy and timing limits. Keep endpoint gating, span caps, and hashed run correlation. Add no span attributes or exception text. ## Verification - Nine new tests passed with real PostgreSQL, real Git, and SDK-exported spans. They cover directory contention, grant-row blocking, all three binding checks, reuse without cloning, dirty files, private-task denial, receipt repair, conflict retry, endpoint-off behavior, and parent IDs. - Twenty earlier unrelated authorization decisions do not exhaust the admission detail budget. - Existing authorization and trust suites: 105 tests passed. Existing timing/context suites: 26 tests passed. Existing repository-access tests: 6 tests passed. - `pnpm -r typecheck` and `pnpm build` passed on commit `5a986d8c86644dfb7ee39b3b832e2d7c86532598`. - `pnpm test:run` could not complete locally. The temporary five-connection cluster returned `sorry, too many clients already` in chat fixtures, and the run was interrupted. It is not a pass. The complete standard CI workflow passed on the same commit, including all 12 server shards and all three chat shards. - Final current-head checks: 54 success, two neutral, two skipped, zero pending or failing. The [CI aggregate](https://github.com/paperclipai/paperclip/actions/runs/39522187940/job/118971261973) passed. [Greptile](https://github.com/paperclipai/paperclip/pull/17040#issuecomment-6305066520) and [Paperclip Review](https://github.com/paperclipai/paperclip/pull/17040#issuecomment-6305017182) both assessed this exact commit at 5/5 with no actionable findings or unresolved threads. - `node scripts/check-module-boundaries.mjs` and `git diff --check` passed. - This macOS host exhausted its SysV semaphore quota. Database tests used a temporary command-scoped initdb preload with five connections, fewer workers, and a larger lock table. No repository test infrastructure or shared host settings changed. Initial skipped tests, focused-fixture failures, and clock-order assertion failures were corrected and rerun; they are not counted as passes. The complete local run remained connection-limited and was interrupted. - Local SDK export proves parentage and privacy. Hosted Honeycomb ingestion remains unverified until deployment. No deployment occurred. ## Risks Span recording requires an operator-configured OTLP endpoint. Each fixed detail name remains capped at 16 occurrences per attempt. Later operations still execute under their enclosing phase. Timing wrappers preserve the existing operations and their errors. No migration is required. The change does not itself reduce admission latency. ## Model Used OpenAI Codex, based on GPT-6, with tool use, code execution, and delegated agents. The exact hosted model identifier and context-window size are not exposed in 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> |
||
|
DottaandPaperclip
|
023f4a5d0c |
fix: preserve repository cache negotiation refs (#16687)
## Thinking Path > - Paperclip manages agent work and its execution environments. > - A fresh repository task needs an independent checkout before its runner can start. > - The object cache from #16487 should reduce repeated network transfer. > - The cache kept Git objects but dropped remote-tracking refs during its bare clone. > - Git needs those tips to negotiate which objects it already has. > - This pull request preserves the tips in a versioned mirror cache and records explicit cache outcomes. > - The benefit is less repeated transfer and clearer evidence when a cache does not help. ## Linked Issues or Issue Description Refs #16487 and #16454. No duplicate cache-negotiation fix was found in the public PR search. **What happened?** A valid object cache could make a fresh clone slower. A controlled staging probe took 45.87 seconds and received 184.39 MB even though the reference already held the objects. An earlier probe took 52.33 seconds. An uncached clone took 20.94 seconds. `git clone --bare` copied the pristine checkout's objects, local branches, and tags, but omitted its remote-tracking refs. The source had about 1,200 remote branches. Git could not use the missing tips to negotiate those objects. **Expected behavior** Keep the pristine reference's negotiation tips. Each task must still authenticate online, use the current remote refs, and end with independent Git objects. **Steps to reproduce** 1. Clone a repository with many remote branches into a fresh checkout. 2. Publish a reference with `git clone --bare --no-hardlinks` from that checkout. 3. Clone the remote with `--reference-if-able` and `--dissociate`. 4. Compare with a reference that retains the remote-tracking tips. Use Git Trace2 and count received pack bytes. **Paperclip version or commit** The staging runtime was `27ab902d3db2770ada79c9f42a9c06e1d346f34b`, with Git 2.47.3. The cache implementation was unchanged from #16487. **Deployment mode** Docker on a persistent application volume. The probe used the application file owner and the public Paperclip repository. It did not start agent tasks or alter the application cache. ## What Changed - Publish pristine object caches with `--mirror` to retain remote-tracking tips. - Put cache format version 2 in the key and receipt. The first checkout builds a new generation. Existing generations stay under the current storage limits. - Keep the existing online authentication, current remote refs, credential and company isolation, independent objects, quotas, and fallback behavior. - Add closed Honeycomb cache outcomes and a boolean that states whether a reference was supplied to clone. - Document that a lookup hit or supplied reference does not prove Git avoided transfer. ## Verification - Controlled staging component experiment: a disposable reference copy held exactly the same 210,057 objects as the original. It added 1,187 advertised branch tips whose commits were already in those objects. - Before: 45.87 seconds and 184.39 MB received, one sample in this comparison. - After: 12.20 and 14.58 seconds; median 13.39 seconds. Received pack sizes were 1.99 and 1.95 MB. - All comparison clones used the same current remote HEAD and ended without an alternates file. - The remaining repack took 9.72–9.81 seconds. Checkout file writing took 0.86–0.96 seconds. - The staging experiment restored tips on a disposable cache copy. It did not deploy this patch or measure complete agent latency. - Confirmed both staging scratch locations were empty afterward. The original cache timestamps stayed unchanged. The temporary operator SSH key was revoked and its local files removed. - TypeScript 5.9.3 transpilation reported zero syntax diagnostics across the four changed TypeScript files. `git diff --check` passed. - The new negotiation regression fails with the old `--bare` publisher and passes with the mirror publisher. - Focused Vitest suites: 26 tests passed across repository-object-cache and execution-preparation-timing. These cover independent objects, current refs, unavailable/revoked remotes, scope separation, invalid storage, tracing isolation, and endpoint-off behavior. - The first full focused run passed 24 tests but failed its existing localhost HTTP fixture because the sandbox denied `listen` on `127.0.0.1`. The rerun with local loopback permission passed all 26 tests. - Server semantic typechecking passed with `tsc -p server/tsconfig.json --noEmit` after building the required workspace declarations. Initial attempts lacked generated dependency outputs; the completed dependency build and rerun passed. - A full application build was not run locally. CI passed: 55 checks passed and two optional checks were skipped. Greptile is 5/5 on the current head, with both findings resolved. ## Risks - Version 2 causes one cold cache rebuild. Old cache generations retain their disk usage, and the existing quota can prevent publication. - A cache still needs a full repack to produce an independent task checkout. This PR does not remove that cost. - The component experiment establishes the negotiation effect on staging. It does not establish an end-to-end agent speedup or the under-30-second goal. - The new trace fields remain behind the existing OTLP gate. Their values are closed enums or a boolean. They export no refs, paths, URLs, cache keys, credentials, or error text. ## Model Used OpenAI Codex, GPT-6 family, with code and tool use. The exact serving model ID and context window are not exposed in 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> |
||
|
DottaandPaperclip
|
ce0eff7d1f |
perf: reduce fresh workspace clone and sandbox staging waits (#16487)
## Thinking Path > - Paperclip manages AI agents and their work. > - Each run needs a private workspace and safe persistence. > - Fresh isolated repository sources download upstream Git history again. > - Sandbox staging also verifies an optional seed cache before it uploads the workspace. > - That verification holds a company lock during a full extraction and hash scan. > - This pull request reuses pristine Git objects through an authenticated remote clone. > - It also overlaps seed verification with upload and moves verification outside the publication lock. > - The benefit is less repeated transfer and less serial preparation work, with the same task isolation and recovery contract. ## Linked Issues or Issue Description **What existing behavior does this improve?** Fresh task-private repository preparation and inbound sandbox workspace staging. **Current behavior** Each isolated remote source clones the full upstream history. Optional seed verification then extracts and hashes another full tree before upload. One workspace's verification can hold the company cache lock while other workspaces wait. **Proposed behavior** Use a company and credential scoped cache of pristine upstream objects for GitHub HTTPS sources. Each task still clones from the current remote. It disconnects from the cache before publication. Run seed verification alongside upload. Hold the company lock only for final capacity checks and publication. **Reason and benefit** These operations lie on the workspace preparation path. Removing repeated Git pack transfer and serial verification can reduce startup time. Local measurements below cover these boundaries. They do not establish hosted agent latency. Related: #16425, #16454, #16398, #16461, #16468, #16473. The cache inspection and durable folder proposal in #13376 covers separate product work. ## What Changed - Add a pristine upstream Git object cache for fresh isolated GitHub HTTPS sources. Key it by company, repository and current credential fingerprint. Do not store credentials. - Keep online clone authorization and current remote refs. Use `--reference-if-able --dissociate` so task objects remain independent. A reference failure retries once online without the cache and resolves credentials again. - Cap cache storage at 64 entries and 4 GiB per company, with 2 GiB per entry. Skip contended publication. Bound priming by a cooperative five-second budget and a 4,096-entry scan allowance. Keep the existing free-space reserve. - Overlap optional seed verification and required workspace upload. Settle both before returning or deleting temporary files. Recheck quotas and company directory identity under the publication lock. - Keep the independent fresh upstream reproducibility proof, task commits, dirty source copies, ignored files, nested repositories and durable recovery behavior. - Suppress optional Git server-info generation during dissociation so generated storage passes the existing safe-expiration checks. This uses [the supported Git setting](https://git-scm.com/docs/git-config/2.43.0#Documentation/git-config.txt-repackupdateServerInfo). - Add bounded timing phases, regression coverage and two repeatable local benchmarks. No first-party Telemetry or dependency changes. ## Verification - `pnpm -r typecheck`: passed. - `pnpm build`: passed. - Affected adapter suites: 240 tests passed. Native workspace persistence/history and tracing: 33 tests passed. Server checkout/isolation/tracing batch: 53 tests passed. Final object-cache authority, path and capacity suite: 12 tests passed. These batches overlap. - Cache and expiration/restoration lifecycle batch: 34 tests passed. A direct generated-storage check covers cached clones. - Real HTTP Git fixture: a matching cache does not bypass revoked remote access. Other fixtures cover current refs/tags, hidden cached commits, independent objects after cache removal, corruption fallback, credential/company scope, quota limits and publication contention. - Initial restricted runs hit SQLite write and Unix socket permission failures. The adapter batch passed with disposable writable instance state and socket access. The first typecheck rejected the test use of `Promise.withResolvers`; typed deferred gates now use the supported library contract. - Repository benchmark on final head `9047368bf3b7fa2061e00d698b8b51cf4ae1da32`: five alternating samples, real local `file://` Git transport, synthetic 32 MiB history, no injected latency. Median 1143.8 ms to 240.5 ms; ranges 1085.7–1486.6 ms and 188.0–254.8 ms. Received Git pack bytes fell from 33,569,788 to zero. Initial cache priming took 1208.4 ms and is reported separately. - Sandbox benchmark, five alternating samples per mode on 10,000 files with real packing, extraction and verification: with a controlled 3-second transfer delay, median 9.450 s to 7.620 s (ranges 9.280–12.115 s and 7.296–7.983 s). With no injected delay, median 6.808 s to 6.450 s (ranges 6.255–7.029 s and 5.709–6.902 s). The no-delay ranges overlap; local copying competes with verification for disk. Both modes verify final tree contents, ignored-file exclusion and Git HEAD. These are synthetic local fixtures, not hosted agent runs. - Reproduce with `pnpm --filter @paperclipai/server exec tsx ../scripts/benchmark-repository-object-cache.ts` and `pnpm --filter @paperclipai/server exec tsx ../scripts/benchmark-workspace-staging.ts --baseline-ref=307eb16c0c --transfer-delay-ms=3000`. - Prior head `af766bc820` received Greptile 5/5 with no findings. Its E2E shard 7 failed in `plan-task-fixture` cleanup: company deletion raced an `activity_log` foreign key. That fixture has no provider run or Git workspace. The failure is retained here; final-head CI must pass. The corrected corruption fixture now forces a pack and verifies a real damaged archive. Current head `9047368bf3b7fa2061e00d698b8b51cf4ae1da32` has Greptile 5/5 with no review threads or findings. Final head has 55 passing checks and two optional Storybook skips. [The first workspace-shard attempt](https://github.com/paperclipai/paperclip/actions/runs/38802441857/job/116655485350) passed all 1,778 cases, then reported a late `EPIPE` from unchanged `server-utils.ts` while the ACP reuse test was active. That case mocks workspace preparation and passed locally in isolation. [Its single retry on the same commit](https://github.com/paperclipai/paperclip/actions/runs/38802441857/job/116659728034) passed, as did the dependent verification gate. The earlier E2E cleanup fixture passed on this final head. Both initial failures remain recorded here. A full local suite was not duplicated; the affected boundaries were run locally and the complete suite runs in CI. ## Risks - First cache population adds local copying. Cold clones, credential rotation, capacity exhaustion and a corrupt cache retain the online clone path. No eviction is added. Cached data is bounded, but many companies can each consume their allowance. - Individual filesystem calls and cleanup can exceed the cooperative priming budget. Failed reference clones make one additional online attempt. - Parallel verification and upload compete for disk and CPU. The local no-delay benchmark shows a smaller gain than a slow-transfer case. Verification still checks contents before cache publication, and required upload failures remain failures. - Local fixtures do not prove hosted throughput or the 15–30-second end-to-end target. The fresh checkout still writes its working tree, and model startup, copyback and other preparation costs remain. ## Model Used OpenAI Codex, GPT-6 family. The exact serving model ID and context window are not exposed in this session. Used reasoning, source editing, shell tools and 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 and contains no internal task identifiers - [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> |
||
|
DottaandPaperclip
|
0e58f29eb7 |
fix(observability): trace pre-dispatch controller waits (#16515)
Measure usage startup, MCP handoff, dispatch admission, and budget reservation with the existing endpoint-gated tracer. Preserve outer attempt context across asynchronous provider dispatch.
Validated with 135 focused tests, full typecheck/build, fresh full CI/E2E, Superagent, and Apex 5/5 on
|
||
|
DottaandPaperclip
|
25b16e65fc |
fix(observability): detail workspace and connector preparation (#16454)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - The server prepares a repository, sandbox workspace, and connector access before an agent can reply. > - Existing traces measure large preparation phases but leave important work inside those phases unclear. > - Repository source materialization and Browser Use discovery can do more work than their callers suggest. > - New detail must preserve access checks, run ownership, and the existing work order. > - This pull request adds fixed OpenTelemetry spans inside those operations. > - Operators can use one deployed trace to select the next performance fix. ## Linked Issues or Issue Description Refs: #16419, #16425, #16428. **What existing behavior does this improve?** Startup traces for server workspace preparation, native sandbox synchronization, and connector assignment resolution. **Current behavior** Workspace and connector aggregate spans can take many seconds. The trace does not separate private source cloning, seed reuse, archive work, progress persistence, or each connector resolver. Browser Use resource selection builds the full authorized tool catalog. The configuration and native backend can each resolve connector assignments. **Proposed behavior** Show the awaited work as fixed child spans. Separate discovery queue wait from catalog construction. Preserve the initiating run context when another run releases a scheduler slot. Add no new lifecycle event writes for workspace detail callbacks. **Reason and benefit** Operators need evidence about the hosted critical path before another performance change. These spans distinguish filesystem work, provider transfer, authorization, transaction entry, locks, and reporting overhead. **Breaking changes** None. Authorization, query order, transaction ownership, synchronization, existing progress events, and task results retain their contracts. The change adds opt-in Observability spans under the existing OTLP endpoint gate. ## What Changed - Trace private source directory setup, Git credentials, clone, requested ref, source identity, and checkout publication. - Trace sandbox source capture, seed cache/replay, shallow clone and archive, overlay copy/archive, verification, upload, warm stamps, and descriptor persistence. - Separate existing awaited progress reporting from useful workspace work. Keep new detail callbacks trace-only. - Trace each connector resolver and Browser Use catalog/grant work. Split discovery queueing, descriptor sources, and policy evaluation. - Trace Slack authority, credential reads, catalog synchronization, transaction entry/settlement, and advisory locks. - Update the real filesystem staging trace expectations and the Honeycomb investigation guide. - Add a real scheduler regression for a queued job released by another run. Assert run ownership, parent spans, and exactly one completed queue span. ## Verification - Passed: server and adapter-utils TypeScript checks (`pnpm --filter @paperclipai/server exec tsc --noEmit` and `pnpm --filter @paperclipai/adapter-utils exec tsc --noEmit`). - Passed: 252 checks across 9 focused files. These cover real filesystem staging, native workspace synchronization, trace ownership, endpoint gating, and workspace preparation. - Passed after review: 26 trace-context, tracing-failure, and discovery-scheduler checks across 3 files (these overlap the first run). The new queued-discovery regression fails when context binding is removed and passes after restoration. An initial submillisecond timestamp assertion was replaced with explicit span-completion order to avoid SDK clock precision assumptions. - Passed on implementation commit `0e68968a4bd61001893d833f281f12bb110edbe1`: `pnpm -r typecheck` and `pnpm build`. The later commit updates only an engine test expectation. - The first CI workspace lane found an old engine trace-tree expectation. It expected the earlier span set. The updated assertion now checks the new detail spans, their parents, and single completion. Its targeted executor check passes. - The full test suite, build, release registry, and end-to-end gates pass in CI on final head `40e37d89ef679109d60a96bb5731a0616efeaa43`: 55 passing checks and 2 optional Storybook skips. The full suite was not duplicated locally after the focused run. - Greptile reviewed the final head at 5/5. No unresolved review threads remain. - After deployment, filter Honeycomb by the hashed run ID and verify `service.version`. Inspect `heartbeat.prepare.workspace`, `runtime_selection.workspace_sync`, and the connector assignment parents. The guide in `doc/observability.md` explains each new family and the nested intervals. - No hosted run has used this new instrumentation yet. Local checks cannot establish hosted latency improvement. ## Risks - Tracing adds spans when an operator configures an OTLP endpoint. Detail remains capped at 16 spans per phase per execution attempt. - Parent and child durations overlap. The documentation explains how to avoid counting them twice. - A discovery job can start in another run's scheduler continuation. The bound callback preserves the initiating async context. - New spans export fixed names and the existing hashed run ID. They add no paths, repository URLs, connector resource IDs, credentials, tool schemas, or exception text. ## Model Used OpenAI Codex, GPT-6 family, with reasoning, tool use, and code execution. The exact served model ID and context-window size are not exposed in 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> |
||
|
DottaandPaperclip
|
57fa6b358d |
fix(observability): separate workspace finalization and copyback spans (#16428)
Give native workspace finalization its own span name. Keep copyback and commit as separate child measurements, with an exported-span regression that preserves the workspace operation result. Validation: 40 runtime/tracing tests and server typecheck passed locally. Final-head CI, E2E and security checks passed. Apex review: 5/5, zero findings and no unresolved threads. Co-Authored-By: Paperclip <noreply@paperclip.ing> |
||
|
DottaandPaperclip
|
cccde7122b |
Trace execution preparation and durable response persistence (#16419)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - The server prepares an execution environment before it starts an agent. > - Operators need to distinguish preparation, provider wait, and response persistence. > - Existing preparation phases start separate traces, and the native session trace ends before the final response is saved. > - This pull request adds an execution-attempt trace and detailed spans at the existing work boundaries. > - The benefit is a measurable startup and settlement path without changing execution policy. ## Linked Issues or Issue Description **What existing behavior does this improve?** Operator-configured OpenTelemetry tracing for agent preparation, startup, and durable response storage. **Subsystem affected** Server orchestration, native Runner startup, and the Honeycomb run link. **Current behavior** Preparation phases start independent roots. Workspace lock waits, authorization groups, runtime asset work, provider-pack fallback, and the gap after provider completion are hard to attribute. **Proposed behavior** Use a server-owned async context for each execution attempt. Parent nested work to that attempt. Link the separately sampled native session because its ingress and queue history can begin earlier. Record transaction entry, lock waits, grouped authorization, asset preparation, artifact fallback, copyback, and response commit. **Reason and benefit** Operators can identify expensive work from recorded boundaries instead of inferring the cause from elapsed time. Nested and concurrent durations remain visible and must not be added together. **Breaking changes** No API or execution-policy change. New Observability span names and closed count attributes are documented. The existing local run-performance schema also records the additional provider-pack and harness measurements. Refs: #16178, #16205, #16330. ## What Changed - Add `task.execution_attempt` with same-run async ownership, independent root sampling, and native-session links. - Measure workspace transaction entry and settlement, advisory and row waits, source scans, reader groups, notifications, and writes. - Measure runtime selection and each concurrent instructions, skills, and MCP branch. Separate manifest, cache, materialization, and verification work. - Measure provider-pack discovery, verified reuse or upload, final verification, and remote harness state checks. - Measure workspace settlement, accounting, liveness, response authorization, comment commit, and publication enqueue. - Preserve the silent endpoint-off path, including when another library has registered a tracer. Bound repeated detail spans to 16 per name per attempt. - Extend the Honeycomb run link and document the actual coverage and remaining gaps. ## Verification - After resolving the conflict with retained-reader batching on `b98cdaea531c8447637cd4dcc19c61b3f5f41844`, all 61 targeted query, authorization, workspace-preparation and trace-boundary tests passed. The server typecheck passed. Reader tests still bound query count, avoid loading retained run payloads and enforce fresh project grants on every publication attempt. - Before that base integration, the tracing/exporter group passed 855 tests with 2 optional-SDK skips across 10 files and no inherited OTLP endpoint. It covers real PostgreSQL waits/rollback, durable response commit, assembled heartbeat preparation, concurrent parentage, sampling and tracer failure isolation. Native workspace/history tests passed 23 cases. These results remain supporting evidence for the unchanged tracing paths. - Full workspace typecheck and build passed during implementation. Full sharded CI, release validation, E2E and Apex 5/5 passed on prior head `8cebad80f3856dbcff6f3a6d5d34880ff77a20f2`. Fresh CI, security and Apex review are required on the current head before merge. - `pnpm check:token-gates` passed. No styles changed. - Earlier CI runs exposed exact-source-expression assertions changed by wrappers, missing run hashes on transaction-boundary spans, and sampling fixtures without an enabled endpoint. The fixes preserve the calls, add the declared hash and span cap, and enable tracing explicitly in the sampling fixtures. Focused repairs passed; the full prior-head CI passed after those repairs. Earlier failed run: https://github.com/paperclipai/paperclip/actions/runs/38729131366. - The full local `pnpm test:run` was stopped after the base advanced during that process. Its mixed-version results are not final-head evidence. A full local-suite pass is not claimed. - The old-head Superagent scan stalled at 35%. The current head has a fresh scan; no security pass is inferred from the old pending result. - No deployed Honeycomb acceptance is claimed by this code change. ## Risks - Enabled tracing creates more spans. Repeated detail spans are bounded, and inventories use group spans. - The attempt and native session remain linked traces. Historical queue time can precede the attempt, and nested intervals overlap. - A committed comment proves durable server storage. Publication enqueue and realtime dispatch do not prove external delivery or browser paint. - New custom attributes use only the existing run hash and validated file/byte counts. No SQL, paths, permissions, content, or exception text is added. ## Model Used OpenAI GPT-6 through Codex. The serving model ID and context window size are not exposed in this session. The agent used reasoning, tools, code editing, 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 --------- Co-authored-by: Paperclip <noreply@paperclip.ing> |
||
|
DottaandPaperclip
|
ebceb2a744 |
trace(server): time workspace, native setup and startup recovery (#16330)
## Thinking Path > - Paperclip manages agents and their work. > - The controller prepares workspaces before it starts a Runner. > - A slow run can spend 30–80 seconds inside the combined workspace phase. > - Startup recovery can also delay queued work for minutes. > - Native runs also have gaps before Runner startup and after the provider result. > - Existing traces do not identify the awaited steps inside those intervals. > - This change adds closed timing boundaries around the existing operations. > - Operators can locate the wait without changing execution or exposing content. ## Linked Issues or Issue Description **What existing behavior does this improve?** Operator-configured OpenTelemetry diagnosis of controller preparation and startup recovery. Related: #16178, #16236 and #16285 introduced the existing preparation timing and sampling contract. #16306 and #16312 reduce separate configuration and copyback work. **Current behavior** The workspace interval combines source resolution, trust checks, provisioning, database persistence, binding and repository admission. Startup awaits accounting and native recovery before queued work resumes, but neither has a separate duration. Native backend setup and post-turn cleanup also contain unmeasured awaits. CPU profiles and a fast database probe cannot identify which await is slow. **Proposed behavior** Record workspace steps and native controller setup and cleanup with the existing hashed run ID. Record startup phases as independent, operator-sampled root spans without custom attributes. Keep the same operation order, results, failures and authorization. **Reason and benefit** These boundaries distinguish source work, database writes, repository admission and restart recovery. This is a diagnosis change. It does not claim a response-time improvement or a solved latency budget. ## What Changed - Add 14 closed workspace step names around 16 existing awaits. Repeated updates and bindings produce separate spans. - Time startup tool reviews, gateway sessions, delivery-worker readiness, accounting, plugin readiness, native recovery, hot restart, orphan reaping, retries, queued work and goal recovery. - Time native GitHub setup, wake binding, backend construction, connector assignments, MCP discovery, tool authority, companion selection and binary resolution. Time selected post-turn GitHub, coordinator and instruction cleanup awaits. - Move preparation timing to a pure shared module. Keep the existing heartbeat import as an alias. Both heartbeat and native timing skip tracer lookup when the endpoint is unset or blank, including when another library registered a provider. - Tracing lookup, callback, status and end faults cannot repeat work or replace its value or error. - Add real PostgreSQL workspace timing coverage and silent no-op/fault controls. Update the tracing contract and paired run maps. - Keep span attributes closed. Startup spans add none; workspace steps use only the existing run hash. No commands, paths, returned records or exception messages are recorded. ## Verification - Diagnostic-gap baseline: the new real workspace test fails against the unchanged workspace implementation because the expected step spans are absent. It passes with the wrappers. - Final affected coverage: **804 distinct cases pass** across seven complete suites: 698 native-executor cases, 39 real PostgreSQL workspace cases, 61 timing/context/startup cases, and six execution-configuration integration cases. The final native and unit invocations pass with no selected skips. The workspace suite passed in the initial six-file invocation on the same final workspace source. - Preserve the initial six-file attempt as failed: **7 failed and 789 passed**. All seven errors occurred during embedded PostgreSQL startup. The log reports local macOS shared-memory exhaustion. A fresh 15-case database group passes, followed by the complete 698-case native suite. These runs overlap; do not add their counts. - The new native timing assertion fails against unchanged native code because the expected spans are absent. It passes with the wrappers and retains the credential-isolation assertions. - Full `pnpm -r typecheck` and `pnpm build`: passed on the final production source. Earlier nullable-closure errors are fixed with erased assertions inside existing guards. The failed attempts remain recorded. - Normalized AST comparison: removing only timing wrappers, timing imports/local handles and erased assertions restores the original workspace, startup and native executor programs. This checks call order and policy expressions; it is not live acceptance. - The expanded head first failed two configuration integration assertions because their test providers had no endpoint. Configure the endpoint explicitly in those two tracing tests, and restore it after each test. All six cases then pass with the original credential cleanup, redaction and exact span assertions. Production is unchanged by this repair. - The first PR head failed one unchanged attachment test at its 15-second deadline in CI. The complete 32-case attachment suite passes locally with that deadline. Do not label the failed CI attempt green. The expanded head needs fresh full CI. - `node scripts/check-maps.mjs`: passed, 35 areas and 176 capability IDs. `node scripts/storybook-catalog.mjs`: 283 files and 2,728 stories. Structural checks only; no browser, play or Storybook build claim. - `git diff --check`: passed. No added prose comments, lockfile changes or new 1,000-line crossings. - A duplicate full local test run was not started. Fresh exact-head full CI and E2E pass on `15a683bb2419960f611584f8c07cdcfc9ce3af0b`. All 56 check runs are successful or optional skipped. Superagent and legacy Snyk pass. Completed Apex review is 5/5 on this same head, with zero unresolved review threads. Commands: ```sh pnpm exec vitest run --project @paperclipai/server --maxWorkers=1 --fileParallelism=false server/src/services/native-runtime/native-session-executor.test.ts server/src/services/execution-preparation-timing.test.ts server/src/startup-recovery-timing.test.ts server/src/services/heartbeat/preparation-timing.test.ts server/src/services/heartbeat/preparation-timing-context.test.ts server/src/services/heartbeat/workspace-preparation.test.ts pnpm -r typecheck pnpm build node scripts/check-maps.mjs node scripts/storybook-catalog.mjs ``` ## Risks - Configured operators receive more spans. They still follow the configured sampler and exporter. - Startup spans measure the awaited phase, including remaining readiness waits. They do not measure earlier worker construction or a provider turn. - Workspace and native setup intervals overlap their containing phases. Do not add them to the parent duration. Native timing covers named awaits, not every statement or every error path. - These spans are not yet deployed or tested against a live latency run. No first-party Telemetry event, run-log write, schema or API changes are included. ## Model Used OpenAI GPT-6 through Codex, with tool use and code execution. The exact served variant and context-window size are not exposed to this task. ## 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> |