mirror of
https://github.com/paperclipai/paperclip.git
synced 2026-11-12 06:10:44 +01:00
42bc238aa5a1208cabecba76773e0500fcd33959
548
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 DottaandClaude Opus 5.5
|
42bc238aa5 |
fix(runner): keep a restart continuation when the lost turn's terminal is delivered again (#20196)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - When a restart loses a native run's active turn, the run records it as `provider_turn_lost_on_restore` and continues in a same-conversation continuation turn > - #20118 taught the event consumer to skip a stale restart-lost terminal of another turn once a continuation is running. It applied only when the checkpoint already recorded the interruption > - When the checkpoint instead names the lost turn as active (the drain hit right after the first turn was submitted), the consumer starts the continuation on the first lost terminal. If the same lost terminal is then delivered again, it was taken as the run's terminal and failed the run > - This pull request applies the skip to any recovered run once its continuation has started. For a checkpointed interruption it still waits for the startup decision first > - The benefit is that a turn lost on restore always continues, however often its lost terminal is delivered ## Linked Issues or Issue Description Refs #20118 **What happened?** Local restart-chaos K13 regression storm w-20261109T055902-s9402 (master |
||
|
Nicky LeachandClaude Opus 5.5
|
9046884811 |
perf(native-runtime): claim a parked Runner while the follow-up run prepares (#20195)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - Paperclip Runner executes native agent runs. A host can park an idle
Runner (release it and freeze it) between runs, and resume it for the
next run on the same task.
> - A follow-up run that reuses a parked Runner must claim it first. The
claim is a bootstrap ticket, a checkpoint flush, and the plugin resume
that thaws the Runner.
> - Core made the claim only inside `attachRun`, after all of its
per-run preparation. The claim does not depend on that preparation, so
the two ran in series.
> - This pull request starts the claim while Core still prepares the
run. `attachRun` then waits for the claim that is already in progress.
> - The benefit is a shorter start for every follow-up run on a parked
Runner. The model below gives about 285 ms per run at P50.
## Linked Issues or Issue Description
No public issue exists. The issue is described inline.
**What happened?**
A follow-up run that reuses a parked Runner claims the Runner only in
`attachRun`. Core calls `attachRun` after backend setup: runner
environment, assigned MCP tools, tool authority and the plugin task
owner. The claim waits for all of this, but it uses none of it.
**Expected behavior**
The claim runs at the same time as backend setup. Everything that needs
the new run's tool authority stays at attach.
**Steps to reproduce**
1. Run a Codex or Claude task on a host that parks Runners between runs.
2. Let the run finish and the Runner park.
3. Send a follow-up. Before this change, the plugin resume starts only
after backend setup ends.
**Paperclip version or commit**
`master` at
|
||
|
Nicky LeachandClaude Opus 5.5
|
92a18bb0f3 |
perf(runner): restart the ACPX provider when a turn settles (#20190)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - Paperclip Runner drives Claude through the ACPX sidecar. The sidecar's MCP endpoint lasts for the whole session, so it cannot tell which turn sent a late tool callback. > - Because of this, the Runner reaps the sidecar and provider of a generation that ran a turn before it releases call-ID tombstones and rotates receipt epochs. Today this restart runs at the start of the next turn. > - Every follow-up turn in a run pays that restart on its critical path: about 1.5 s with real Claude. #20145 moves the restart to park, but a second turn in the same run, after the first turn that follows an attach, still restarts cold. > - The restart does not have to wait for the next turn. When a turn settles and the host has its terminal result, the Runner is idle. > - This pull request restarts the provider at that point. The next turn, or the next park, then finds a fresh generation. > - The benefit: same-run follow-up turns start without a restart, park does no restart after a settled turn, and the late-callback guarantee is unchanged. ## Linked Issues or Issue Description Refs #20145, which moved the restart to park. This pull request builds on it. No public issue exists, so the problem is described here. **Subsystem affected** Paperclip Runner, ACPX provider (Claude). **Current behavior** `AcpxProviderSession::start_turn_timed` restarts the provider before any turn whose generation already ran a turn. A follow-up turn in the same run waits for `session.suspend`, sidecar shutdown, a new sidecar, `initialize`, `session.open` and `run.attach` before `turn.start`. Since #20145, park does the restart instead, but only for the first turn after an attach. **Proposed behavior** When a turn settles and the controller has ACKed every committed event, the Runner restarts the provider. The next `turn.start` uses the fresh generation and only rotates receipts. Park reuses the fresh generation and does no second restart. **Reason and benefit** The restart leaves the turn's critical path, both for same-run follow-ups and for the turns after a park. The order of reap and rotation stays the same, so a late callback from the reaped generation still cannot reach a later turn. ## What Changed - **Runner loop** (`durable/runner.rs`): new `CommandExecutor::prepare_next_turn` hook, a no-op by default. The runner calls it after the idle provider poll, only when no control frame is waiting and the controller has ACKed every committed event (`acked_source_seq >= highest_source_seq`). So the restart always runs after the terminal result reaches the host. - **ACPX turn-end restart** (`acpx_park_preparation.rs`, `acpx_provider_generation.rs`): - `restart_session_after_turn` restarts a Claude session that has a completed turn, nothing pending, no buffered events, a dormant goal and its whole event tail ACKed. It first settles an abandoned park observation. If that settlement reads more events, it skips this round, and the next idle round restarts after those events are ACKed. - `AcpxProviderSession::restart_after_turn` does the same suspend → reap → start → bootstrap → identity check as the turn-start restart. It records the reaped process id in `restarted_after_turn_from`. This field is the explicit "fresh because of a turn-end restart" marker. - The session does not poll a fresh generation until its next turn or rebind, the same as a prepared one. Its startup output is published with the next turn, as today. Polling it early would add events after the host computed its park cut. - **Next turn** (`acpx_provider_session.rs`): `start_turn_timed` restarts only when the generation ran a turn. When `restarted_after_turn_from` is set, it does the post-restart receipt and ledger rotation without a second restart. `session.reconciled` reports the reaped process as `previousProcessId`. - **Park**: `prepare_next_generation` restarts (with the 20 s park deadline) only if no turn-end restart ran. Otherwise it marks the generation for `run.attach` rebind and returns `false`, so the runner does not ask for a second park proof. The park-time receipt rotation moves to the next turn start, where the turn-end path does it too. - **Restart failure**: the executor sets the same state that a failed turn-start restart leaves (`closed`, no session, provider exit confirmed). The next `turn.start` fails with the same `ACPX turn start failed closed: …` error that the turn-start restart gives today. - **Fake sidecar**: new `pending-input` fault, which opens an input request in a turn that then completes. ### Races | Race | Handling | | --- | --- | | A turn or steer arrives during the turn-end restart | The runner loop is single-threaded. The command waits in the socket until the restart returns, then starts on the fresh generation without a restart. Only one generation is live. | | A turn arrives before the hook runs (for example, in the same read as the ACK) | Control frames go first, so the turn restarts at its start, as today. The hook then restarts after that turn. | | Park arrives during the turn-end restart | Host control is read at the top of the loop, so park waits. Park then reuses the fresh generation. No event is published in between, so the host's park cut still matches. | | Park arrives before the hook runs | Park restarts with the 20 s deadline, as in #20145. The hook then finds a fresh generation and does nothing. | | Eviction, `runner.suspend` or retire after the restart | They reap the fresh generation through the ordinary paths. | | The restart fails | The session is `closed`. The next turn fails with the same error and the same durable state as a failed turn-start restart. Park returns conflict, as it does for a closed session today. | | Cancellation | `run.cancel`, `turn.interrupt` and `turn.stop` after the restart return `already_settled`, as after any settled turn. A cancellation that retires the runtime closes the session, so no restart runs. | | Interrupted, cancelled or failed turn | `completed_turn_observed` is false, so no turn-end restart runs. The next turn restarts at its start, as today. | | Pending permission or input requests at the terminal | The reducer ends them at the terminal, so the restart runs. A later `request.resolve` fails with `no active turn`, with or without the restart. | | Active session goal at the terminal | No restart, because the goal can keep working after its prompt. | ### Removed as redundant, and what stays - **Removed:** the park-time restart and dormant verify when a turn-end restart already made the generation fresh. The runner's `host_park_ready` proves dormancy just before `prepare_for_park`, and nothing runs in between. - **Removed:** the runner's second park proof in that case, because the provider did not change. - **Removed:** the park-time receipt and ledger rotation. The next turn start does it once, for both paths. - **Kept: the turn-start restart.** Without it, a turn queued before the hook, a turn after an interrupted or cancelled turn, a turn after an active goal, and any non-Claude ACPX turn would run in the generation that ran the previous turn. Tests cover the queued-turn and interrupted cases. - **Kept: the park-time restart.** Without it, a park that arrives before the hook would leave the next run to restart at attach. A test covers this. - **Kept: the settlement of an abandoned park observation.** Without it, the restart drops events that the old generation sent before the abandoned response. A test covers this, and it fails with the settlement removed. - **Kept: the goal-dormancy check.** Without it, the restart kills an active goal's autonomous work. A test covers this, and it fails with the check removed. - **Kept: the ACK gate.** Without it, the restart can run before the host has the terminal result. A runner test covers this. - **Kept: deferred polling of the fresh generation.** Without it, its startup event lands after the host computed its park cut, and the park conflicts. - **Kept: the kept failure message.** Without it, the next turn fails with `ACPX session is not open`, and the restart error is lost. ## Verification - `cargo test -p paperclip-runner-core --lib` and every `tests/acpx_*.rs`, `native_provider_backend` and `durable_recovery` suite: all pass (debug). - New tests (`tests/support/acpx_turn_end_restart.rs`, run from `acpx_park_preparation`) prove these points: - A settled turn reaps the old pid before it spawns the new one, and no predecessor is still live. A same-run follow-up then sends only `turn.start` and reports the reaped pid in `session.reconciled`. - Park after a turn-end restart sends no sidecar command and returns `false`. Attach rebinds the same pid. The run's second turn restarts at its end, and the third turn starts without a restart. - The reaped generation's late callback never reaches the next turn. - A turn that arrives before the hook restarts at its start. A park before the hook restarts once. - A failed turn-end restart gives the same next-turn error, retry error, durable state and process count as a failed turn-start restart. - Cancellation after the restart is `already_settled`. `runner.suspend` reaps the fresh generation. - An active goal skips the restart. An input request open at the terminal behaves the same with or without the restart. - `acpx_provider_turns`: a session-level turn-end restart drops the old generation's late `issues.delete` callback, which reuses the previous turn's call ID. The next turn then admits that call ID from the fresh provider. Interrupted and retired turns do not restart at turn end. - `acpx_park_readiness`: a turn-end-restarted generation is park-ready. An abandoned park observation tail is delivered before the restart. - `durable::runner`: the hook runs only after the terminal is ACKed and when control is idle. It never runs during ACK debt, a waiting control frame or a disconnect. - Server integration tests (`plugin-runner-release-resume`, `plugin-runner-eviction` and `plugin-runner-task-warm`) with this branch's Runner: 49 passed. The provider process counts do not change: each released run still adds one process start, now at turn end instead of at park. - `cargo fmt --check`: clean. - Fake-sidecar timing (release build, 7 runs, medians): | | before (`master` with #20145) | this PR | | --- | --- | --- | | same-run follow-up `turn.start` → `turn.started` | 51.5 ms | 22.0 ms | | second turn after an attach, `turn.start` → `turn.started` | 270.3 ms | 23.5 ms | | turn-end restart, after the terminal is ACKed | — | 29.0 ms | | park preparation after a settled turn | 34.1 ms | 0.03 ms | - Real Claude: not run for this pull request. With real Claude, the restart took about 1.5 s on the turn-start path, and this change takes it off that path. That number is not measured on this branch. ## Risks - A run that ends after one turn without a park now starts one extra sidecar that no turn uses. The Claude CLI itself starts only at the first prompt, so the cost is the sidecar's start and bootstrap. - The restart blocks the Runner loop for its length (about 1.5 s with real Claude), as the turn-start restart does today. It has no deadline, as before. A park that arrives during it waits, inside the host's 30 s park wait. - The fresh generation is not polled until its next turn. If it exits while idle, the next turn finds out at `turn.start` and fails closed. Park finds out at its observation. - The turn-end restart is for Claude only, the same as park preparation. Other ACPX agents keep restarting at turn start. - If the Runner crashes during the restart, the durable state still says `session_open`. Recovery is the same as for a crash while idle. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), in Claude Code with tool use (shell, file edits, local 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 - [ ] 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> |
||
|
Nicky LeachandClaude Opus 5.5
|
2d93413085 |
perf(runner): prepare the next Claude provider generation at park (#20145)
## Thinking Path
> - Paperclip runs agents through Paperclip Runner, and a host can park
an idle Runner between runs and resume it for the next run on the same
task.
> - For Claude (ACPX), a follow-up run on a parked Runner pays about
1.5–2.3 s at `run.attach` before the provider can start the turn. The
attachment closes the Claude sidecar and spawns a new one: initialize,
`session.open` with `session/load`, `run.attach`, then a dormant verify.
> - The respawn is required. The tool-bridge MCP endpoint lasts for the
whole session and can't tell which turn a late callback came from, so a
new turn must not run in a process that ran an earlier one.
> - Nothing requires the respawn to happen at the *start* of the next
run. A turn that ends 2 s later is a better experience than one that
starts 2 s later.
> - This pull request does the respawn while parking. The next run's
`run.attach` then only rebinds the run id on a fresh, never-used
generation.
> - Follow-up Claude runs on a parked Runner reach the provider about
1.5–2 s sooner, and the late-callback guarantees are unchanged.
## Linked Issues or Issue Description
No public issue exists. The issue is described inline.
### What happened?
On a hosted deployment with parked Runners, Claude follow-up runs
reached the provider about 1.8 s later than Codex follow-ups.
`runner.session.resume` medians were 5.2 s for Claude and 3.4 s for
Codex. The extra time is the ACPX respawn inside `run.attach`. Codex
keeps its process in place.
### Expected behavior
The respawn happens while the Runner parks, so the next run starts on a
prepared generation.
### Steps to reproduce
1. Run a Claude (ACPX) task on a host that parks Runners between runs.
2. Let the run finish and the Runner park.
3. Send a follow-up. `run.attach` closes the provider and starts,
initializes, opens and attaches a new sidecar before `turn.start`. The
Claude CLI itself starts at the first prompt on either path.
### Paperclip version or commit
`master` at
|
||
|
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
|
cedf335c98 |
fix(native): a failed terminal followed by an active restart continuation is not a permanent failure (#20149)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - When a restart loses a native run's active turn, the run records that turn as failed with `provider_turn_lost_on_restore` and continues on a same-conversation continuation turn > - Restart recovery blocks a run as `provider_checkpoint_permanently_failed` when its checkpoint carries a failed run terminal and no pending restart interruption > - While a continuation is running, the checkpoint keeps the lost turn's failed run terminal and names the continuation as the active turn. The pending-interruption check returns nothing once a turn is active > - So a second restart during the continuation read the run as permanently failed, blocked replay, and needed the board > - This pull request treats that failed terminal as the restart loss it records whenever an active turn follows a restart-lost terminal turn > - The benefit is that runs survive repeated restarts during their continuations, including M1 parents and their subtasks ## Linked Issues or Issue Description Refs #20148 **What happened?** Local restart-chaos M1 K4/K6 rerun w-20261109T003126-s9701 (master |
||
|
DottaandClaude Opus 5.5
|
b43cccbb08 |
fix(runner): keep a restart continuation when a stale earlier continuation is reported lost (#20118)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - When a restart loses a native run's active turn, the runner reports it as `provider_turn_lost_on_restore` and the controller starts one same-conversation continuation. A later restart that loses the continuation is answered the same way > - With back-to-back restarts, the controller's checkpoint can miss a continuation that the runner already accepted. On the next restore the controller starts a fresh continuation, and the restored runner then reports the unrecorded earlier continuation as lost > - The event consumer treated that stale lost terminal as the run's terminal, so the run failed while its new continuation was already running > - Recovery then refused to replay the failed run, and the task needed a Retry. In an M1 parent/subtask task, the parent's Retry succeeded but the parent stayed `blocked` on that subtask > - This pull request makes the consumer skip a restart-lost terminal for any turn other than the continuation it started > - The benefit is that back-to-back restarts no longer turn a recoverable lost turn into a failed run that needs Retry ## Linked Issues or Issue Description Refs #20107 **What happened?** Restart-chaos K14 wave w-20261108T213600-s9803 (local, an upgrade between builds plus repeated restarts) at |
||
|
DottaandClaude Opus 5.5
|
412d49e48d |
fix(runner): resume an ACPX turn interrupted by a deploy instead of refusing the launch digest (#20112)
## Thinking Path
- The rule for upgrade-safe saved state: work saved by the old build
must resume under the new build.
- The K14 upgrade-mid-run chaos wave (`838e9bf3d2` → `f789f3e2af`) broke
that rule for Claude. The same heartbeat run resumed after the upgrade
restart, and the runner's resume probe (`runner.drain`) failed with
`ACPX durable launch profile digest does not match runner startup`
(`provider_state_incompatible`). t03 and t04's Claude subtask both
needed a board Retry. Codex crossed the upgrade cleanly.
- The ACPX launch profile digest covers the provider-pack authority and
the sidecar artifact hashes, so every deploy changes it.
- #18879 already re-binds a suspended session to the new digest. A
deploy that restarts the runner mid-turn leaves the same run with an
active turn, or with an open session, which is the state a same-build
lost-turn restore leaves behind. Neither was re-bound.
- A same-build restart already recovers both states safely: it reopens
an open session, and recovers an active turn through
`restore_conversation_after_lost_turn`, which is guarded by the
provider-lifetime fence. An upgraded build should take the same path.
## Linked Issues or Issue Description
Refs #18879
**What happened?**
A deploy restarted the runner while a Claude (ACPX) task was mid-run.
When the same run resumed under the new build, restore refused the saved
state with `provider_state_incompatible`. The task stopped and needed a
manual Retry, which started a fresh provider session and dropped the
conversation.
**Steps to reproduce**
1. Start a Claude task on build A and let it reach an active turn.
2. Upgrade to build B, which ships a new provider pack or sidecar build,
with a graceful restart while the turn runs.
3. The same run's resume fails with `ACPX durable launch profile digest
does not match runner startup`.
**Expected behavior**
The resumed run recovers exactly as it would after a same-build restart.
The conversation is restored, or the lost turn is closed through the
normal recovery path, with no Retry.
## What Changed
-
`packages/paperclip-runner/runner/crates/runner-core/src/acpx_provider_backend.rs`:
- `rebinds_settled_session` now also re-binds the saved digest for an
**interrupted turn or open session of the same heartbeat run**
(`rebinds_interrupted_turn`).
- The re-bind requires exactly the same-build recovery preconditions: no
accepted semantic result, no pending runtime request, no unconfirmed
provider exit, and a dormant goal. It also keeps the existing
requirements: no candidate admission, and the same sidecar command and
arguments.
- Recovery then runs unchanged. The provider-lifetime fence still
decides whether the conversation is restored or the lost turn is closed.
- **Relocated installs:** an upgrade can install the same launch files
under a new directory, as an `npx` versioned cache, a new image layout
or a new checkout does. The K14 harness moves between worktrees. The
re-bind now accepts a launch of the same shape: the same executable and
sidecar file names, and identical non-path arguments. It then records
this startup's runner-owned paths in the saved descriptor. A renamed
executable or a changed argument stays strict.
- An active turn or open session saved by a *different* run stays
strict, as before. So do suspended sessions with undelivered events, and
every qualified-release binding (ACPX, agent server and command digest).
- Tests (Rust, `acpx_provider_backend`):
-
`interrupted_turn_of_the_same_run_rebinds_to_the_upgraded_runner_launch_profile`:
red on master. An old-digest saved state with an active turn restores,
the digest is re-bound, and identity and command digest are kept.
-
`open_session_of_the_same_run_rebinds_to_the_upgraded_runner_launch_profile`:
restore gets past the saved-state check, to provider start.
- `saved_sessions_rebind_to_an_upgraded_build_installed_at_a_new_path`:
red on master and on the first fix, with the production error. It covers
an interrupted turn of the same run and a settled session attached by
the next run, both moved to a new install directory.
-
`interrupted_turn_after_upgrade_stays_strict_when_lost_turn_recovery_is_unsafe`:
each unsafe guard still fails with `provider_state_incompatible` and
leaves the saved state untouched.
- The existing strict tests for other runs are unchanged and still pass.
- `packages/paperclip-runner/docs/durable-recovery.md`: documents the
extended re-bind.
- `tests/acpx_warm_attachment.rs`: the end-to-end test
`restores_a_turn_lost_to_a_runner_upgrade_under_the_new_launch_profile`,
from the Phase 2 session's parallel fix. It fails on master with the
production error and passes here.
## Verification
- `cargo test -p paperclip-runner-core --lib acpx_provider_backend`: 44
passed.
- The full `paperclip-runner-core` lib suite: 585 passed. The 3 failures
are timing-based `task_memory` kill-monitor tests on macOS; that file is
untouched (last changed 5 days ago). CI runs them on Linux.
- `cargo fmt --check`: clean.
- Live: the first K14 rerun, at
|
||
|
DottaandClaude Opus 5.5
|
15b987d9b8 |
fix(runner): a finish or block waits for held calls before it is judged (#20108)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - Restarts must be invisible to a task: an agent that was working
keeps working after the control plane comes back
> - Restart chaos t05 (Daytona warm wave, codex, K4 hold) ended a task
as blocked even though its document save succeeded
> - #20069 refuses a finish or block that the agent decided before it
saw results that had already returned. Here the results had not returned
yet when the block was checked: after the reattach, the held calls and
the later block are handled at the same time
> - This pull request makes a finish or block wait, for a bounded time,
for earlier calls of its turn that have not returned, and then apply the
same check
> - The benefit is that a controller outage during a Paperclip tool call
no longer turns into a false "blocked" task
## Linked Issues or Issue Description
Refs #19831 (durable ownership plan). Restart chaos wave
`w-20261108T192459-s9501`, task t05 (Daytona warm, codex, C1, K4 hold of
182 s, master
|
||
|
DottaandClaude Opus 5.5
|
a8bbdf237c |
fix(runner): reopen the ACPX session at turn.start when it can be safely reopened (#20110)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - A native Claude ACPX run must survive restarts and suspensions. A
turn stopped for suspension drops the live ACPX session and leaves the
saved lifecycle `prepared`
> - The next `turn.start` checked for a live session before anything
else, so it failed with "ACPX session is not open". That happened even
though the saved session was safe to reopen
> - In restart chaos this made a Claude approval task (G1 t03) need a
Retry after repeated restores
> - This pull request makes `turn.start` reopen a reopenable session
before it starts the turn
> - The benefit is that a suspended or restored Claude run keeps going
on its own, with no Retry
## Linked Issues or Issue Description
Refs #20089
**What happened?**
In a local restart-chaos G1 (approval) wave, Claude ACPX task t03 failed
after repeated restores with `ACPX session is not open` and needed a
Retry. The Retry overwrote the session directory, so the exact sequence
could not be read back. The same error reproduces directly with
`turn.stop` (suspension) followed by `turn.start` on the same session:
the stop clears the live session and sets the lifecycle to `prepared`,
and `turn.start` never reopens it.
**Expected behavior**
When the saved session can be reopened, `turn.start` reopens it, the
same way `session.open` does, and then starts the turn. "Can be
reopened" means: the conversation identity is recorded, the provider
exit is confirmed, the session belongs to this run, and the lifecycle is
`prepared`, `suspended` or `session_open`. A session that was closed on
purpose, or whose provider exit is unconfirmed, still fails.
**Steps to reproduce**
1. `run.prepare`, `session.open`, then `turn.start`.
2. `turn.stop` (suspension stop).
3. `turn.start` again. Before this change, it fails with "ACPX session
is not open".
**Paperclip version or commit**
master at
|
||
|
DottaandClaude Opus 5.5
|
a88e23fa34 |
fix(runner): read an uncommitted turn start as empty history when the event window is complete (#20107)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - A native run must survive repeated control-plane restarts; each
restore reconciles the turn its checkpoint names as active
> - That reconciliation reads the turn's history. When the provider has
no paginated history, the runnerd transport rebuilds it from the
committed runner events
> - After back-to-back hard restarts, the checkpoint can name an
accepted turn whose start was never committed. The runner was killed
after journaling `turn.start` and before the provider emitted anything
> - The rebuild treated every missing start as lost history and failed
with `codex_history_incomplete`, so the task needed a Retry
> - This pull request treats a missing start as empty history when the
committed window is provably complete
> - The benefit is that a run survives back-to-back restores, as it
already survives one
## Linked Issues or Issue Description
Refs #20074
**What happened?**
In restart-chaos wave w-20261108T200620-s9701 (local, K4/K6 + M1,
back-to-back K4 restores), three runs failed on the second restore. Two
were Claude ACPX C1 tasks; the third was an M1 parent with its subtask.
They failed with `native_session_recovery_failed:
HarnessReconciliationError: codex_history_read_failed: thread/items/list
requires supported paginated history; Error: codex_history_incomplete:
requested turn start is outside the retained runner event window`, and
each needed a Retry.
The sequence:
1. The first restore lost the in-flight turn and started a continuation
turn. ACPX echoes the requested turn id, so the checkpoint named it as
active.
2. Six seconds later a second K4 SIGKILL killed the runner. The runner
had journaled that `turn.start` (now `execution_indeterminate`), but the
provider had emitted nothing, so no `turn.started` was committed and the
runner outbox was empty.
3. On the second restore, `thread/items/list` for that turn hit the
guard, although the committed window was complete: source sequences
1–25, nothing evicted.
**Expected behavior**
The second restore treats the never-started turn like the first restore
treated its lost turn. The runner reports it lost on restore, and
recovery starts a continuation with no Retry.
**Steps to reproduce**
1. Have a checkpoint name an accepted turn whose `turn.started` was
never committed. For example, kill the runner after `turn.start` and
before the provider emits anything.
2. Restore the session. Before this change, recovery fails with
`codex_history_incomplete`.
**Paperclip version or commit**
master at
|
||
|
DottaandClaude Opus 5.5
|
a9ae1b69c3 |
fix(runner): let a finish through after a restart lost the provider turn (#20088)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - The Runner relays an agent's Paperclip tool calls and keeps them across control-plane restarts > - #20069 refuses a `paperclip_finish` or `paperclip_block` decided before an earlier returned result reached the agent > - The guard filters by turn id, but each committed event carries the run's one identity turn id, so provider turns are not separated > - When a restart loses a provider turn, a call of that turn gets its result but never its provider completion > - The guard then refuses every later finish of the run, so the run exhausts its retries and the user must click Retry > - This pull request makes a new provider turn end the pending calls of the previous turn > - The benefit is that an agent can finish after a restart that lost its turn ## Linked Issues or Issue Description Refs #20069 **What happened?** A restart chaos storm (local, Codex, A1/C1, K2 drains for 30 minutes, 116 tasks) had one failure. Task t110 needed a user Retry. Its run was drained twice. The first drain lost the provider turn while a `read_agent_instructions` call was pending. After that, all four `paperclip_finish` calls failed with `terminal_call_before_results` and named `read_agent_instructions`. The run ended with `native_session_retry_exhausted`. **Steps to reproduce** 1. Start a Codex run that calls a Paperclip tool. 2. Drain the control plane while that call is pending, so the Runner restores the session and reports `provider_turn_lost_on_restore`. 3. Let the agent continue in the new turn and call `paperclip_finish`. **Expected behavior** The finish is recorded. The lost call belongs to a turn that the agent will not see again. **Actual behavior** The Runner refuses every finish with `terminal_call_before_results`. The run fails and needs a user Retry. ## What Changed - `unseenReturnedToolCalls` forgets the calls it tracked when a `turn.started` event begins a new provider turn. - A new test reproduces the t110 event order: a call in a lost turn, then a new turn that reads, sees its result, and finishes. ## Verification - `npx vitest run src/live/terminal-call-freshness.test.ts` in `packages/paperclip-runner`: the new test fails before the change and passes after it (7/7). - `npx vitest run src/live/runnerd-codex-transport.test.ts -t terminal`: 10 passed. - `tsc --noEmit -p packages/paperclip-runner`: clean. - Evidence: the kept Runner control-plane state of t110 shows the call `exec-0f3d1b90` (generation 1) with a `semantic_tool.result` command but no `tool.execution.completed`, followed by `turn.failed provider_turn_lost_on_restore` and new `turn.started` events. ## Risks - The t03 case of #20069 stays covered: the stale block and the earlier calls are in the same provider turn, with no `turn.started` between them. The existing t03 tests still pass. - If a driver emits `turn.started` inside one provider turn, the guard can miss a stale call of that turn. No current driver does that. ## 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
|
c55af26821 |
fix(runner): treat a replayed start of a turn lost on restore as history (t48) (#20074)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - A native Codex run must survive a control-plane restart and continue its work > - #20057 made recovery adopt an in-flight turn, or start a new turn when the restored runner reports the old turn lost > - One case was not covered. A graceful drain can detach the controller after the runner accepted `turn.start` but before the provider started the turn > - The runner then records the provider's `turn.started` while no controller is attached. After a restart it replays that start, and the Codex session rejects it as an unexpected turn > - This pull request treats that replayed start as history from the earlier generation > - The benefit is that the run starts its continuation turn instead of stopping with `native_event_replay_conflict` ## Linked Issues or Issue Description Refs #20057 **What happened?** A restart chaos wave ended with one failed task (t48, Codex). A graceful SIGTERM suspended the controller 300 ms after the runner accepted `turn.start`. The provider's `turn.started` came after the controller detached. Recovery classified the run as `resume_dead_runner`. The restored runner reported the turn as `provider_turn_lost_on_restore`, which is correct. But it also replayed the undelivered `turn.started`. The Codex session failed with "Provider started an unexpected turn." (`turn_binding_mismatch`), and recovery stopped with `native_event_replay_conflict`. **Expected behavior** Recovery starts a new continuation turn. The replayed start of the lost turn is recorded as history, and the run completes. **Steps to reproduce** 1. Start a Codex native run. 2. Detach the controller gracefully after the runner accepts `turn.start` and before the provider emits `turn/started`. 3. Let the runner record `turn.started` while it is unattached, then kill the runner and the provider. 4. Restore the run. Before this change, the session fails with `turn_binding_mismatch`. **Paperclip version or commit** master at |
||
|
DottaandClaude Opus 5.5
|
f679cf81dc |
fix(runner): refuse a finish or block decided before earlier results reached the agent (#20069)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Restarts must be invisible to a task: an agent that was working keeps working after the control plane comes back > - Restart chaos t03 (Daytona warm wave, K4 holds) ended a task as blocked even though its document was saved > - The cause is in how a terminal decision (paperclip_block or paperclip_finish) is handled after a long controller outage, not in how the Runner journals tool calls > - This pull request refuses a terminal call that the agent made before it had seen results that already returned, so the agent continues with those results > - The benefit is that a controller outage no longer turns into a false "blocked" task ## Linked Issues or Issue Description Restart chaos t03, wave `w-20261108T142906-s9501` (Daytona, codex, A1, three back-to-back K4 holds of 136 s, 252 s and 152 s). Refs #19831 (durable ownership plan). **What happened** The kept instance's run log shows the order of events: 1. The turn is accepted at 14:29:57. The runner then delivers nothing to the controller until 14:39:27. The holds keep the controller down, and the Runner keeps Codex's `item/tool/call` requests until it is reattached. 2. During that outage Codex keeps running. Its `write_document` call stays pending. The model is sampled again about once a minute and says "The save request is also waiting on Paperclip". It then calls `paperclip_block`, with the reason `tool_unavailable`. 3. After the reattach the Runner delivers the backlog in order. `write_document` runs and returns (the document is created: sourceSeq 152). Then the stale block arrives (sourceSeq 247) before Codex has acknowledged the write result: there is no `tool.execution.completed` for it before the block. The task ends blocked. **Codex behavior, checked directly** against the pinned Codex 0.160 app-server. A local probe sent one dynamic tool call and left it unanswered: - Codex does not time out. It re-samples the model at about 60 s and 127 s ("The save call is still waiting to return"). - A reply sent 130 s later is accepted (`durationMs` 130003), and the model finishes ("DONE"). So the reply is not lost. The agent's terminal decision was made without it. **Expected behavior** - A terminal call decided while earlier Paperclip results were still unseen is refused with a retryable error. - The agent continues with those results. In t03 it would see that the document was saved, and finish. ## What Changed - `packages/paperclip-runner/src/live/terminal-call-freshness.ts`: `unseenReturnedToolCalls` reads the turn's committed run events up to the terminal call. It returns each dynamic call that has a returned `tool_result` but no provider `tool.execution.completed`. `staleTerminalCallResult` builds the retryable `terminal_call_before_results` error, which names those calls. - `runnerd-codex-transport.ts` `#handleSemanticToolInput`: when such calls exist, `paperclip_finish` and `paperclip_block` get that error instead of being dispatched. The recovery path after a restart uses the same handler. A long-running call with no result yet does not count. - Crash point `runner.semantic_tool.after_input_before_dispatch`, just before the controller dispatches a dynamic tool call. A crash-point hold (`--hold-ms`, `--crash-point-hold-ms`; a K4-style receipt) keeps the controller down after the self-kill. Together they reproduce t03 on Daytona. ## Verification - Red/green: `cd packages/paperclip-runner && npx vitest run src/live/runnerd-codex-transport.test.ts -t "refuses a terminal call decided"`. - The case `seen: false` failed on master: the handler was called and the stale block was dispatched. It passes with the fix. - The case `seen: true` (the agent saw the result first) dispatches normally, both before and after. - `npx vitest run src/live/runnerd-codex-transport.test.ts`: 289 passed. `src/live/terminal-call-freshness.test.ts`: 5 passed. - `npx tsc --noEmit -p packages/paperclip-runner` passes, and `node scripts/typecheck-test-projects.mjs tests/restart-chaos` passes. - Shared registry: 15 passed. Call-site test: 3 passed. Chaos runner tests: 12 passed. - Codex probe (local, about 2.5 min per run): reproduced the "still waiting" re-sampling and the accepted late reply described above. - Not yet run: the Daytona crash-point repro with a hold. It needs the shared harness slot, and I'll add the result here. The fix is controller TypeScript, which is not an input of the Daytona image, so the current image still matches. ## Risks - A refused terminal call costs the agent one more turn step. It applies only when a result returned that the agent had not seen, which in practice happens only after a controller outage. - The check reads committed run events. If the store evicted the earlier call's events, it finds nothing and keeps today's behavior. - Upgrade-safe run state: no persisted shape changes. The check reads events the Runner already commits. A run saved by the previous release resumes with the same history and the same answer. - ACPX and other providers emit no dynamic `tool.execution.*` events, so the check is a no-op for them. ## 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
|
66a2cbab98 |
fix(runner): adopt the in-flight Codex turn after a restart lands just after turn.start (#20057)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - A native Codex run survives a server restart by resuming its durable provider session and adopting the turn that was in flight > - The adoption rule (RC-59) only adopts a turn named by the latest `turn.start` command's accepted result > - Two gaps made it fail. A restart just after `turn.start` leaves that command `execution_indeterminate` (the runner died after journaling it). Separately, the rule demanded that Codex echo the requested turn id, which it never does. > - The restored session then rejected Codex's own in-flight turn as unexpected, and the user's Retry failed on the first click > - This pull request adopts exactly the turn the runner reconciled for that ambiguous start, and accepts Codex's own turn ids > - The benefit is that a restart landing right after a turn starts no longer costs the user a second Retry ## Linked Issues or Issue Description Found by the restart-chaos campaign: wave `w-20261108T134237-s9401`, task t77 (Codex C1), Retry run `5358f1cb`. Related: #19852 (fast follow-up turns), RC-59 adoption. No duplicate PR was found. **What happened?** The Retry run submitted its first turn at 14:05:25.171. A K2 SIGTERM followed at .690 (graceful drain, then `resume_dead_runner`). - The runner had journaled the `turn.start` and then died, so on restore it settled the command as `failed / execution_indeterminate`. - The runner's restored Codex session reconciled the ambiguous start to the live replacement turn (`session.reconciled`, `activeProviderTurnId`). - Codex then sent `turn/started` for that turn. The restored driver had no active or pending turn, because `controllerAcceptedTurnId()` was null, and failed with `turn_binding_mismatch`. - The run ended `native_event_replay_conflict`, and the user needed a second Retry. **Expected behavior** The restored session adopts the runner-reconciled in-flight turn and finishes it. It does not treat it as a replay conflict. **Steps to reproduce** 1. Start a native Codex run. 2. Send SIGTERM to the server about 0.5 s after the first `turn.start`, before runnerd confirms it. 3. Let the server restart and resume the run (`resume_dead_runner`). 4. Before this PR, the run fails with "Provider started an unexpected turn." **Paperclip version or commit** `master` before this PR. ## What Changed - `controllerAcceptedTurnStartId(commands, options)` (runnerd-codex-transport.ts): - **Codex and OpenCode results.** These providers assign their own turn ids, so an accepted result never echoes the requested `turn_lab_` id, and RC-59 adoption never matched a real Codex result. For them, the accepted turn is now adopted only while it is also the runner's live active turn in its authenticated `session.snapshot`. A turn the restored runner reports lost has no live turn, so it keeps the established restart-continuation path. ACPX still requires the echoed id. - **Indeterminate start.** When the latest `turn.start` (after the latest attach) failed with `execution_indeterminate`, it adopts the runner's live active turn. Runnerd reports one only after it reconciles an ambiguous start to a distinct replacement turn. - The transport records the snapshot's active turn at `thread/read`, which `recover()` reads before it evaluates adoption. - The driver's existing RC-59 guards still apply: the turn must be in provider history, be the only active turn, and not already be terminal. ## Verification - Connected restart test, `runnerd-restart-continuation.test.ts` "adopts the in-flight turn when a restart lands just after turn.start (t77)". It uses the real runnerd and fake Codex: - the runner and provider die after journaling a started turn; - runnerd's own restore settles the command `execution_indeterminate`; - recovery adopts `provider-turn-1` with no second `turn/start`. - It fails without the fix: nothing is adopted. - Unit cases for `controllerAcceptedTurnStartId`: - a Codex-assigned id is adopted only while it is the live turn; - an indeterminate start is adopted with the reconciled turn; - it fails closed with no live turn, an ordinary failed start, a later attachment, or an ACPX id mismatch. - Regression guarded: "starts a new turn when the restored runner reports a lost turn the checkpoint never bound" still passes. - `runnerd-codex-transport`, `codex-app-server-driver.recovery`, `runnerd-codex-transport-adoption`, `runnerd-codex-transport-cold-restore`, `runnerd-restart-continuation` and `runnerd-controller-outage`: 383/383. Runner TypeScript typecheck is clean. - Live: the chaos campaign will run the `runner.turn_start.*` crash points (Codex A1/C1) and a Codex C1 K2 storm, and post the result here. ## Risks - **Previous-release story.** No persisted shape changes. Commands journaled by the previous release carry forward and are read with the new rule. An indeterminate start that the runner did not reconcile to an active turn still fails closed, as before. - **Wider Codex adoption.** Codex adoption now actually fires. If runnerd ever returned an accepted result that names a turn other than the one this command started, the driver would adopt it. Runnerd binds the result to the exact command id, so this is the same trust the rule already placed in ACPX. ## 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
|
946c098bf3 |
fix(opencode): bind the attached run's contract after a restored launch (#20001)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - Native OpenCode agents run through runnerd and a small OpenCode proxy. The proxy checks each `paperclip_finish` claim against the run's completion contract. > - Each follow-up run has its own contract revision. #19874 lets runnerd keep the OpenCode process across a contract change, and sends the new contract on `turn/start`. > - When runnerd starts again for a follow-up, it restores the saved session first. The restored proxy launches with the saved run's contract, but its environment already names the new run. > - The proxy applied the `turn/start` contract only when the run id changed. Here the run id did not change, so the proxy kept the saved contract and refused every correct finish claim. > - This pull request makes the proxy rebind whenever the turn's contract differs from the one its session holds. The stale-revision error also names the expected revision. > - The benefit: a queued-comment follow-up finishes on its first try, and an agent that sends a stale revision can repair it. ## Linked Issues or Issue Description Refs #19874 (warm runs carry their own completion contract). **What happened** A user sent a follow-up comment while an OpenCode agent was working. After the first run finished, Paperclip started a continuation run for the queued comment. The agent updated the plan and called `paperclip_finish` with the current revision `2`. The tool refused it twice with "Semantic result completion contract revision does not match". The error did not name the expected revision. The agent searched for it for 3 minutes, and cleanup cancelled the run. Root cause: runnerd restores a saved `session_open` session before it applies `run.attach` (`provider_backend.rs`: "execute() restores the durable provider before dispatching run.attach"). The restore launches the OpenCode proxy with the saved contract (revision `1`). Runnerd's environment already names the new run, so the proxy's `PAPERCLIP_RUN_ID` is the new run. `attach_run_in_place` keeps this provider for a changed contract (#19874) and sends `paperclipCompletionContract` revision `2` on `turn/start`. The proxy compared only the run id. The run id matched, so the proxy skipped `attachRun` and kept revision `1`. **Expected behavior** The continuation's first `paperclip_finish` with the current revision is accepted. If a claim has a stale revision, the error names the expected revision. **Steps to reproduce** 1. Use a native OpenCode agent with warm follow-ups (the default). 2. Start a task. While the agent works, post a follow-up comment, so that it queues. 3. Let the first run finish. Paperclip starts a continuation run with completion revision `2` and restores the saved OpenCode session. 4. The agent calls `paperclip_finish` with `contractRevision: "2"`. 5. Before this fix, the tool refuses it with "Semantic result completion contract revision does not match". The chat eval `steering.discard-queued-message` (runner-opencode) shows this in run events: `session.resumed` carries revision `2`, and both `paperclip_finish` results fail with the mismatch. ## What Changed - `opencode-app-server-proxy.ts`: record the contract that the session opened with. On `turn/start`, call `attachRun` when the run id changes or when the supplied `paperclipCompletionContract` differs from the session's contract. - `opencode-proxy-task-envelope.ts`: add `sameOpenCodeProxyCompletionContract`. - `opencode-server-driver.ts`: the stale-revision error now ends with `Use contractRevision "<expected>".`, like the criteria error. - Tests: a real-proxy regression for the restored launch, a driver test for the new error text, and a `repair-revision` mode in the fake OpenCode fixture. ## Verification - `pnpm exec vitest run src/cli/opencode-proxy-input.test.ts src/drivers/opencode/opencode-server-driver.test.ts src/cli/opencode-app-server-proxy.test.ts src/cli/opencode-proxy-thread-read.test.ts` in `packages/paperclip-runner`: 88 passed. - Red/green: without the source changes, both new tests fail. `binds the current contract when a restored launch already names the attached run` never sees the finish call reach the controller. `names the expected revision in a repairable stale-revision tool error` does not find the expected revision in the error. - `npx tsc --noEmit -p packages/paperclip-runner`: no errors in the changed files. (My worktree is missing a link for `@paperclipai/paperclip-eval-kernel`, which causes unrelated errors in `src/eval/workflow-harness.ts`.) - End-to-end: the chat eval `steering.discard-queued-message` (runner-opencode) with this fix cherry-picked. The run link goes in a comment below. ## Risks - Previous-release story (`rules/upgrade-safe-run-state.mdc`): this change does not change a persisted shape. The regression test uses the saved-state shape: launch parameters from the saved run, environment and `turn/start` from the new run. A runnerd that does not send `paperclipCompletionContract` behaves as before, so the proxy keeps its launch contract. The proxy and runnerd ship in the same runner artifact. - The proxy can now call `attachRun` for the same run id when the contract changes. `attachRun` resets the per-run result state. This occurs only at `turn/start`, when no turn is active, and only when the contract is different. A result for the earlier contract is not valid for the new contract. - The error text changed. No code parses 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`) 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> |
||
|
Nicky Leach
|
e7e870cfc4 | fix(native): decide warm and adopted Runner process reuse from one key (#20005) | ||
|
DottaandClaude Opus 5.5
|
65f41593d1 |
fix(runner): tell agents a skill found in a directory is imported, not re-created (#19983)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Agents can add skills to the company library in two ways: author new content with `create_skill`, or import an existing skill so its source is kept > - #18717 told agents in the `create_skill` contract to import "an existing skill" instead of copying it > - An agent read "existing" as "already in the company library" > - It searched the library, found no match, and saved a managed copy of a SKILL.md from a vetted local directory > - The copy has no source locator, so update, audit and reset flows cannot follow the source > - This pull request says in the same contract that a SKILL.md in a local directory, repository, URL or catalog is an existing skill even when the library has no match > - The benefit is that agents import such skills and keep their provenance ## Linked Issues or Issue Description Refs #18717 **What happened** A person asked the CEO agent: "Find a useful accessibility-review skill and add it to our company." The conversation pointed to a vetted local directory with a SKILL.md. The agent called `search_skills`, found no match, and wrote in its reasoning "No existing accessibility skill. Now create it via paperclip_create_skill". It saved a `managed` skill with `sourceLocator: null`. It did not call the import API that the `create_skill` description recommends. **Expected behavior** The agent imports the skill from its directory through `POST /api/companies/{companyId}/skills/import`. The saved skill is `local_path` with the directory as its source locator. **Steps to reproduce** 1. Put a SKILL.md in a directory that a company project workspace allows. 2. Give the directory path to the agent in chat, and ask it to add the skill to the company. 3. Look at the saved skill. Before this change, the agent can save a `managed` copy without a source locator. ## What Changed - Add one sentence to the `create_skill` description in `packages/paperclip-runner/src/protocol-actions/create-skill.ts`: a SKILL.md in a local directory, repository, URL or catalog is an existing skill, even when the company library has no match, so import it instead of copying its content. - Regenerate the semantic action catalog, the semantic tool contracts, the protocol manifest digest and the seeded catalog digest. - Extend the existing contract test so that the discovered, live and scenario descriptions all contain the new rule. ## Verification - `pnpm exec vitest run src/semantic-tools/paperclip-semantic-tools.test.ts -t "distinguishes authored skills"` in `packages/paperclip-runner` passes. The new assertion fails on the parent, because the parent description does not contain the sentence. - `node scripts/generate-semantic-action-catalog.mjs --check`, `node scripts/generate-protocol-manifest.mjs --check` and `node scripts/generate-semantic-contracts.mjs --check` pass. - End-to-end chat eval `skills.find-and-install` (runner-opencode, local), with this commit cherry-picked onto the eval branch: before, run 42448994428 failed because the agent saved a managed copy. After, run 42450246200 passed. The agent quoted the new sentence in its reasoning, imported the directory through the import API, and the saved skill is `local_path` with the vetted directory as its source locator. ## Risks - Low risk. The change is model-facing text only. It does not change permissions, routes or tool inputs. - Agents that author truly new skills still use `create_skill`. The sentence applies only when a SKILL.md already exists somewhere. ## Model Used - Anthropic Claude Opus 5.5 (`claude-opus-5-5`), 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> |
||
|
DottaandClaude Opus 5.5
|
c0c26188db |
fix(runner): stamp event generations and record an earlier generation's events as history (#19960)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Native runs execute through the Paperclip Runner, which turns provider output (Codex, OpenCode, Claude through ACPX) into durable runner events > - In the restart chaos campaign, events from an earlier turn or runner generation failed the current turn about ten times (RC-15, 20, 20b, 20c, 33, 33b, 59 and 61) > - #19718, #19722, #19772 and #19783 each special-cased one variant, and other variants still fail the run > - The durable-ownership plan (section G, invariant 6) asks for one rule: every event carries its generation and turn, and an earlier generation's events are recorded as history and never fail the current turn > - This pull request stamps the runner generation on every event and applies one stale-event rule in the ACPX provider state, the Codex driver (which Runnerd also uses for OpenCode) and the pending `turn.start` check > - The benefit is that a late report about a settled or restored turn no longer stops a healthy turn, while events for a turn the runner never saw still fail closed ## Linked Issues or Issue Description Refs #19718, #19722, #19772, #19783 **What happened?** A runner or provider can report a turn after that turn settled: a resumed provider reports the lost turn's usage or terminal, an ACPX sidecar emits a last delta after the turn terminal, or a held failure of the previous turn sits in front of a new `turn.start`. Several of these paths fail the current turn: ACPX `AcpxEventScope` rejects any event for a non-active turn ("named a stale turn"), the Codex driver fails `turn_binding_mismatch` for a settled turn's late item between turns, and `runnerdPendingTurnStartFailure` reports any held non-recoverable failure, whatever turn it names. Runner events carried no generation. **Expected behavior** An event for a turn that is not active, and that an earlier generation accepted or this generation already settled, is recorded as history and never fails the current turn. An event for a turn the runner never saw still fails closed. **Steps to reproduce** ACPX: settle turn 1, start turn 2, and deliver a late `runtime.event` for turn 1. Before this change, the provider state returns an error and the run fails. ## What Changed - Runner (`durable/state.rs`, `durable/runner.rs`): every event payload carries `generation`, the runner process generation that emitted it. A runner process that recovers state starts the next generation (`begin_recovered_generation`); it is saved with the next normal write, so recovery adds no write before the first turn. Loading stays pure. Old saved state loads as generation 0. Outbox events keep their original generation. - PRP event schemas v1–v3 declare the optional `generation` property; generated validators and the manifest are regenerated. - ACPX (`acpx_event_scope.rs`, `acpx_provider_state.rs`): an event that names a settled turn of this run becomes a bounded `stale_generation_event` diagnostic (32 per state) and changes no state. Unknown turns still fail. - Codex driver (`codex-session-state.ts`, `codex-session-notifications.ts`): one predicate, `isHistoryTurn()`: the turn is not active, and it was restored from the checkpoint (an earlier generation) or settled in this generation. The classifier's stale-turn branch, the resume-usage branch, restored-terminal replay and the main fail point (`notificationNamesActiveTurn`) all use it. A terminal for an earlier generation's turn is history even when it disagrees with the checkpoint. A terminal that contradicts a turn this same generation settled is still a `conflicting_turn_terminal` integrity failure, and a provider that starts a settled turn again is still rejected. - Transport (`runnerd-codex-transport.ts`): a pending `turn.start` ignores a held failure that names another provider turn. - `DURABLE_TRANSPORT.md` documents the stamp and the rule. ## Verification - `cargo test -p paperclip-runner-core`: new `records_late_events_for_a_settled_turn_as_history` (red before), `stamps_every_event_with_the_runner_process_generation`; two existing ACPX tests now assert "recorded as history, no state change" instead of "rejected". Durable tests: 112 passed. `cargo fmt` clean. - `codex-app-server-driver.stale-generation.test.ts` (new, 9 tests): late item, delta, terminal and workspace change for a settled turn during and before the next turn, an earlier generation's conflicting terminal after recovery, and an unknown turn that still fails. Red on master: 8 of 9 (the during-turn cases on master already dropped the event, with a different diagnostic; the between-turn and recovery cases failed the session). - `runnerd-pending-turn-start-failure.test.ts` (new): an earlier turn's held failure no longer fails the pending start; the pending turn's own failure and a failed run terminal still do. - `vitest run src/drivers/codex src/live/runnerd-codex-transport`: 787 passed; 2 Codex executable-resolution tests timed out under host load and pass when run alone. - `tsc -p tsconfig.json` for the runner package, `check:protocol-types`, `check:replay-goldens`: clean. - Not run: chaos waves (the coordinator schedules them on this host). ## Risks - A late report about a settled turn is now dropped with a diagnostic instead of failing. A provider that keeps working on a settled turn could have its output ignored silently; a provider that starts a settled turn again, or names a turn the runner never saw, still fails closed. - `generation` is a new optional field. Consumers that do not know it ignore it, as the PRP contract requires. - The generation counter is written with the next normal state save, so there is no new write 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
|
fda58d8cc1 |
fix(chat): post the complete final reply, not a 4,000-character run-log preview (#19971)
## 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. > - The runner cuts every completed agent message to a 4,000-character run-log preview. The server then posts that preview as the reply. > - A long answer, for example a list of 501 open tasks, reaches the person cut off with "…[truncated]". > - PR #19959 fixed only the case where the accepted summary starts with the exact preview text. It missed runs where the final prose differs from the summary. > - This pull request keeps the complete final answer in the runner, and makes the server never prefer a preview over a complete source. > - The benefit is that the person receives the agent's complete reply. ## Linked Issues or Issue Description Supersedes #19959. **What happened** A chat agent wrote a final answer of about 4,985 characters (all 501 tasks). The posted chat reply was a 4,000-character preview that ended with "…[truncated]". **Expected behavior** The person receives the agent's complete final message. A preview of a message is never posted when a complete text exists. **Steps to reproduce** 1. Ask a chat agent (OpenCode or Codex) to list a large set of open work, so that its final answer is longer than 4,000 characters. 2. Let the run finish. 3. Read the chat reply. It ends with "…[truncated]" at 4,000 characters. Root cause: - `packages/paperclip-runner/runner/crates/runner-core/src/provider_events.rs`: the codex/OpenCode facade projection of `item/completed` and the ACPX `AssistantMessage` projection cut the final answer with `bounded_text(text, MAX_TEXT_CHARS)` (4,000 characters). - `durable/state.rs` `sanitize_value` caps every string again at 4,096 bytes before the event goes to the server. - `server/src/services/heartbeat-run-summary.ts` `resolveHeartbeatRunResponse` gives an explicit `final` agent message top precedence, so it posts the preview. ## What Changed - Runner (`provider_events.rs`): a completed `final_answer` agent message keeps its complete redacted text, up to a new 32,000-character `MAX_FINAL_AGENT_MESSAGE_CHARS` bound. Progress prose and in-flight items keep the 4,000-character preview. - Runner (`durable/state.rs`): `preserve_final_agent_message_text` keeps the complete redacted text of a completed final agent message when it fits in half of the transport frame. Otherwise the marked preview stays, so a large message never fails the frame check. - Server (`heartbeat-run-summary.ts`): `readCompletedAssistantMessageCandidate` marks a stored preview as `bounded`. The resolver never posts a bounded preview while a complete source exists. It posts a marked adapter final response or the accepted semantic summary instead (`bounded_final_agent_message_replaced_by_semantic_summary`). It posts the marked preview only when no complete text exists (`bounded_final_agent_message_without_complete_source`). This also fixes replies from runners that are already deployed. ## Verification - `cd server && npx vitest run src/__tests__/heartbeat-run-summary.test.ts`: 86 passed. Six new tests: a complete final message longer than 4,000 characters is posted in full; the old exact-prefix case posts the complete accepted answer; a preview whose prose differs from the summary posts the complete summary (also with chat precedence); a marked adapter final response wins over a preview; the marked preview is posted only when nothing complete exists; a stored preview candidate is marked `bounded`. Five of the six fail against the `master` resolver. - `cargo test -p paperclip-runner-core --lib -- final_answer complete_final`: 3 passed. Both new runner tests fail when the fix is reverted. - `cargo test -p paperclip-runner-core --lib`: 581 passed, 3 failed. The failures are `task_memory::tests::monitor_*` (OOM-kill monitor timing tests). They do not touch the changed code. - `cargo test -p paperclip-runner-core --test acpx_event_projection --test acpx_rich_events --test local_integrity_boundary_golden --test acpx_provider_events`: all passed. `cargo fmt --check` passes. - Evidence from the failed eval run: the `item.completed` final `agentMessage` text was exactly 4,000 characters (3,988 retained plus the marker). The `run.result.proposed` summary was complete at 4,985 characters. ## Risks - Larger durable events: a final answer can now be up to 32,000 characters (at most about 128 KB). The outbox keeps the preview when the text exceeds half of the frame size, so the frame check does not fail. - Redaction stays the same. The complete text goes through `redact_sensitive_text_values`, which the preview also used. - The server treats any final message that ends with "…[truncated]" as a preview. If an agent writes that text itself, the server posts the complete accepted summary instead. That summary is also the agent's own complete answer. - The runner part reaches users only after a runner release. The server part fixes replies at once when an accepted summary exists. > 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`), agentic coding with tool use (shell, file edits, 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 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |
||
|
DottaandClaude Opus 5.5
|
f755361ffe |
fix(recovery): deliver a steered message the restart lost before the model read it (#19962)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - A user can steer a running Codex or Pi turn: a comment goes into the live turn instead of waiting for the next run > - The provider acknowledges the steer at once, but the model reads it only at its next step, after the in-flight tool call > - A control-plane restart in that window kills the provider with the message in memory. Restart recovery starts a continuation turn that does not carry the message. The message is lost and no error shows > - The restart chaos harness showed this in 6 of 6 steered runs (Codex, K1, K2 and K6, 2 samples each): steer acknowledged, run succeeded, no reply to the steered message > - This pull request makes restart recovery deliver every acknowledged steer that the model did not read, exactly once > - The benefit is that manual Steer no longer loses messages on restart, and the planned automatic steering can ship on top of it ## Linked Issues or Issue Description **What happened?** A comment steered into a running Codex turn got no reply when a control-plane restart (SIGKILL, SIGTERM or crash loop) happened during the agent's tool call. The run succeeded, and the steered comment had no effect. **Steps to reproduce** 1. Start a Codex task that runs a 40-second command. 2. During the command, post a comment and click Steer. The provider acknowledges it. 3. Restart the control plane before the command ends. 4. Recovery resumes the turn with a continuation. The agent finishes the task and ignores the comment. **Expected behavior** The agent reads and handles the steered comment once, after recovery. **Root cause** The provider keeps acknowledged steering input in memory until its next model step. The restart continuation (`paperclip.native-continuation.v1`) only told the model to continue the current request, so the acknowledged input was gone. ## What Changed - `unreadSteeredMessageIds` (runner package): derives acknowledged-but-unread steers from run-log events in order. A `steering_acknowledgement` item adds the message id (once per id). A `userMessage` item in the same turn reads the oldest unread steer. Codex records each steered input as one `userMessage` item when it enters the conversation. A `userMessage` item in a later turn reads all earlier steers, because the continuation carries them. `turn.completed` reads all steers. - `readUnreadSteeredMessages` (server): reads those facts from `heartbeat_run_events` (only transport-validated events with a `sourceEventId`). It also reads the comment or interaction text for the same company and issue. The run log is the durable record of both "acknowledged" and "read", so no new column is needed. - The restart continuation now includes `steeredMessages` and one instruction to handle them, oldest first. The runtime reads them through a lazy `getUnreadSteeredMessages` option, only when a restart continuation starts. Native recovery wires it for every run. - `queuedInteractionSteeringBody` is extracted, so recovery rebuilds a steered interaction response with the same text the steer sent. - Fake Codex gets `--consume-steer`, which emits the `userMessage` item for an accepted steer, like real Codex. ## Verification - Red/green live test with real runnerd and fake Codex (`runnerd-restart-continuation.test.ts`, "delivers a steered message the restart lost exactly once"): start a held turn, steer it, SIGKILL runner and provider, restore and recover. - "unread": the continuation carries the message exactly once. Without the runtime change this test fails. - "read" (`--consume-steer`): the continuation does not carry it again. - Both: after the continuation completes, a later restart has nothing to deliver. - `steering-delivery.test.ts`: ordering, dedupe, initial-input, later-turn and completed-turn rules. - `unread-steered-messages.test.ts` (embedded Postgres): ignores events without a source id and messages from other issues. It reads interaction text, ignores model-output items as read evidence, and returns nothing after the continuation input. - `pnpm vitest run src/steering-delivery.test.ts src/live/runnerd-restart-continuation.test.ts` (runner) and `pnpm vitest run src/services/native-runtime/unread-steered-messages.test.ts` (server): pass on the rebased revision. - `issue-queued-comments-routes.test.ts` and `native-session-runtime.test.ts -t restart`: pass. `native-session-resume.test.ts` "archives a damaged prior epoch…" fails locally on plain `origin/master` too (local runnerd binary); CI runs it with a fresh build. - Restart chaos (Codex steer mode, K1/K2/K6) runs after this lands in the harness slot. Before: 0 of 6 steered messages answered. ## Risks - Duplicate delivery: a provider that records steered input without a `userMessage` item, or a `userMessage` event lost in the crash window before persistence, makes recovery send the message again. The model then sees it twice. That is safer than losing it, and the instruction asks the model to handle each one once. - Pi (ACPX) steering has no `userMessage` read evidence today, so a Pi steer that the model already read before a restart may arrive again in the continuation. - Only the restart continuation carries the messages. Other recovery paths start after a completed turn, which already read every steer. - No schema or migration change. A session saved by an earlier release resumes normally; its run log already holds the acknowledgement events. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, via Claude Code with tool use (shell, file editing, 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
|
18644a3263 |
feat(recovery): instrument 20 durable write boundaries with crash points (#19938)
## 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 > - #19937 added `crashPoint(name)` and a registry of 20 named boundaries from the Phase 0 inventory > - A name without a call site cannot fire, so the crash-point suite cannot test it > - This pull request puts one call at each registered boundary and adds a test that keeps the registry and the call sites in step > - The benefit is that the on-demand suite (next pull request) can kill the control plane at every known write-ahead gap ## Linked Issues or Issue Description Refs #19831 (plan, Phase 5). Refs #19846 (inventory, "Write-ahead coverage"). Stacked on #19937 (helper and registry). Refs #19861 (ownable kind ids). **Problem** The registered crash points have no call sites. **Expected behavior** Each registered point is called at its gap: after the first durable write or outside effect returns, before the next one starts. Disarmed (always in production) each call is one comparison. **Scope** Call sites and a coverage test only. The suite runner and CI workflow follow. ## What Changed - Server call sites (`@paperclipai/shared/crash-points`): - `sandbox-creation-operations.ts`: around the provider `create` call (intent → provider → lease). - `native-session-executor.ts`: after `native.process_start_requested`, before spawn. - `paperclip-control-plane-port.ts`: after the result-accept transaction, before finalization starts (only when a result was inserted). - `native-run-finalizer.ts`: after `commitNativeStatusDecision`, before the run projection. - `workspace-operations.ts`: `workspace_finalize` only — after the running row, and after `settle()` before the done row. - `environment-runtime.ts`: provider release → lease row; teardown claim → destroy; reusable destroy → lease release; stop-and-retain claim → stop. - `heartbeat.ts`: lease release → runtime services; budget reservation hold → dispatch. - `heartbeat/queue.ts`: run claimed `running` → wake request `claimed`. - `run-cost-accounting.ts`, `usage-receipts.ts`: accounting commit → spool unlink; receipt spool → DB row → unlink. - `packages/paperclip-runner/src/crash-point.ts`: a small copy of the helper for controller code in this package, which does not depend on `@paperclipai/shared` (adding that dependency would change the runner content ID). It arms only `runner.*` names. runnerd never sees the variable: its environment is an allowlist. - Runner call sites: after the runnerd spawn, before its identity is published (both spawn paths); `turn.start` result received → controller journal saved → next send. - `server/src/__tests__/crash-point-call-sites.test.ts`: every registered point has a call site, every call names a registered point, and every `kinds` entry is an ownable kind id of the Phase 1 registry. ## Verification - `cd server && npx vitest run src/__tests__/crash-point-call-sites.test.ts` → 3 passed. - `pnpm --filter @paperclipai/server typecheck` → clean. `pnpm --filter @paperclipai/paperclip-runner exec tsc --noEmit -p .` → clean. - Behavior with nothing armed is unchanged: each call returns at its first comparison (measured in #19937: about 3.7 ns). ## Risks - Low risk. No call changes control flow unless `PAPERCLIP_CRASH_POINT` names it and `NODE_ENV` is not `production`. - Points on the path before `turn.start` (process start, runner spawn) cost nanoseconds, within plan invariant 8. - `usage-receipts.ts`: the persist-then-unlink one-liner became a block with the same statements in the same order. - The runner package change gives the runner a new content ID, as any runner source change does. No persisted run state changes (rules/upgrade-safe-run-state.mdc). ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, via Claude Code with tool use (shell, file edits) and a read-only exploration subagent that mapped the boundaries. ## 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
|
0189fedffc |
fix(native): bound an idle warm local runner whose controller died (#19879)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - A warm runner (`lifecycleMode: warm`) keeps runnerd alive between runs so a follow-up turn starts fast. The follow-up default flip (next PR) makes warm the default for local Codex and OpenCode agents. > - The controller owns a warm runner's idle window. Runnerd ignores the idle timeout itself. > - A graceful controller shutdown suspends idle warm sessions. A controller that dies (SIGKILL, crash) does not. Local runners launch with no process reconnect grace, so the idle runner kept running until its 1-hour connection lease ran out, owned by nobody. > - This pull request gives warm local runners the 7-minute reconnect grace that remote runners already use. > - The benefit: idle runners are reaped by proof within a bounded time, never left orphaned, before warm becomes the default. ## Linked Issues or Issue Description Refs #19874 (warm completion contract), #19846 (TTFT baseline), #19852. **What existing behavior does this improve?** Ownership of an idle warm local runner after its controller dies. **Current behavior** `runnerProcessReconnectGraceMs` is `undefined` for local targets, so runnerd launches without `--reconnect-grace-ms`. An idle warm runner whose controller was killed stays alive until its connection lease (one hour) expires. The restart-chaos harness cannot show this: its K1 kills the controller's whole process tree, runner included. **Proposed behavior** `localRunnerProcessReconnectGraceMs(policy)` returns the remote grace (7 minutes) for a warm policy and `undefined` for per-turn. After the grace, runnerd records `transport_reconnect_grace_exceeded`, stops its provider, keeps durable state and exits. The next run resumes cold. **Reason and benefit** The restart plan requires idle runners to be reaped by proof, never orphaned. Seven minutes is far longer than a controller restart (13–20 s locally) and the long-outage fault holds (3–5 minutes), so a live turn still reattaches. ## What Changed - `native-runner-recovery-contract.ts`: `localRunnerProcessReconnectGraceMs`. - `native-session-executor.ts`: the local runnerd transport passes it as `runnerProcessReconnectGraceMs`. Remote targets keep `null` (lease-owned). - `runnerd-warm-idle-orphan.test.ts` (new): real runnerd and fake Codex. An idle warm runner, after its turn completed, whose controller detaches. - `doc/DEVELOPING.md`: the grace and its bound. ## Verification - `pnpm --filter @paperclipai/paperclip-runner exec vitest run src/live/runnerd-warm-idle-orphan.test.ts`: with a process grace, the idle runner exits with `transport_reconnect_grace_exceeded` and the next transport resumes the same provider session cold. With no grace (`null`, today's local launch), the runner is still alive after the controller left. - `cd server && npx vitest run src/services/native-runtime/native-runner-recovery-contract.test.ts`: warm gets the remote grace; per-turn keeps none; the grace outlasts a controller restart and a long outage. - `tsc --noEmit` passes for `server` and `packages/paperclip-runner`. - Local restart measurement with warm runners (K1/K2, codex, claude, opencode, n = 3 each): every interrupted run recovered, with recovery latency equal or lower than per-turn. The chaos waves with this change are in the default-flip PR. ## Risks - A warm local runner whose controller stays down for more than 7 minutes exits. A turn in progress then resumes through `resume_dead_runner`, the same path as a controller crash that kills the runner. Remote runners already behave this way. - Previous-release story (`rules/upgrade-safe-run-state.mdc`): no persisted state changes. A warm runner launched by the previous release keeps the launch arguments it started with (no grace) until it exits. > 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> |
||
|
Can H. Tartanoglu
|
410aa6d1b4 |
test(runner): resolve provider fixture signals through PATH (#14476)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - Its native runner tests verify exact provider shutdown and recovery. > - Four fixture call sites invoke `/bin/kill` directly. > - NixOS supplies `kill` through PATH, so those tests fail before checking process state. > - This pull request resolves `kill` through PATH, as the neighboring runner tests already do. > - The same stop, probe, and resume assertions can run on non-FHS hosts. ## Linked Issues or Issue Description **What happened?** Four Codex provider tests fail with `No such file or directory` on NixOS. Each failure comes from a test-owned `/bin/kill` invocation. The executable is installed on PATH. **Expected behavior** Tests should probe and signal their disposable provider processes with the installed `kill` command. **Steps to reproduce** On a NixOS host with `kill` on PATH and no `/bin/kill`, run the `codex_provider` Rust integration target. Three `durable_stop_*` cases and `structured_question_round_trips_through_the_normalized_backend` fail. All four failures were reproduced separately on unmodified `e9debd3eac328e6abe87ab89e9ff40ee2bb56b73`. Related: these failures also blocked the broader local qualification of #14474. This fixture fix has no dependency on that runtime change. ## What Changed - Replace four test-only `/bin/kill` executable paths with `kill`. - Retain every signal, PID/process-group operand, assertion, and unwind cleanup. - Match existing PATH lookup in this file and `tests/process_supervisor.rs`. ## Verification Rebased onto master `820c458465` (single commit `ad1fb1043`). - `cd packages/paperclip-runner/runner && cargo test -p paperclip-runner-core --test codex_provider --locked -j4 -- --test-threads=2`: 94 passed, 0 failed, 2 ignored (subprocess helper entries that their parent tests run). The four tests that used `/bin/kill` pass on a host without `/bin/kill`. - No production code or documentation changes. - Hosted CI: `General tests (workspaces-b)` timed out in `packages/db/src/backup-lib.test.ts` and `e2e shard (1/8)` hit `ERR_STREAM_UNABLE_TO_PIPE`. Neither area is touched by this test-only diff. Master pushes currently run the "CI profile skipped" workflow, so there is no master run of those jobs to compare against. ## Risks Low risk. Tests require a working `kill` on their test PATH. Fixture PIDs and arguments stay unchanged. ## Model Used - Anthropic Claude Sonnet 5.5 (`claude-sonnet-5-5`) through Claude Code with tool use and code execution, for the rebase and verification. Original change: OpenAI `gpt-6-astra` through OpenCode. ## 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 |
||
|
DottaandClaude Opus 5.5
|
dcaada5f59 |
fix(native): carry each warm run's own completion contract (#19874)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work.
> - A warm runner (`lifecycleMode: warm`) keeps runnerd and the provider
process alive between runs, so a follow-up message can start fast.
> - Each run has its own completion contract. The revision changes with
each follow-up, and the agent must name that revision when it finishes.
> - On a warm attachment the transport rebuilt the `run.attach` payload
from its constructor options. Those options belong to the first run.
Every later run kept the first run's contract.
> - The OpenCode proxy checks the finish claim locally against that
stale contract. Every warm OpenCode follow-up was refused ("Semantic
result completion contract revision does not match"), and the agent
looped until the run timed out.
> - This pull request carries each run's own contract into the warm
attachment, and lets runnerd adopt it without a cold provider resume.
> - The benefit: warm follow-ups finish correctly, and they keep the
live provider process, which is the reason to be warm.
## Linked Issues or Issue Description
Refs #19852 (startup sub-spans), #19846 (TTFT baseline).
**What existing behavior does this improve?**
Warm cross-run attachment for native Codex and OpenCode agents.
**Current behavior**
In a local measurement wave with `lifecycleMode: warm`, 4 of 4 OpenCode
follow-ups looped for 8+ minutes. Each finish call was refused with
"Semantic result completion contract revision does not match". The saved
`runAttachTemplate` after the attachment still had revision `1`, while
the follow-up turn named revision `2`. Codex hid the defect because its
finish claim goes through the controller's tool path, which uses the
current input.
**Proposed behavior**
The attachment carries the attaching run's contract. Runnerd updates the
contract in place for Codex and OpenCode and keeps the provider process.
The OpenCode proxy receives the current contract with the run identity
on `turn/start`.
**Reason and benefit**
Warm follow-ups are the largest TTFT lever in the baseline. A
correctness-only fix (cold-resume the provider when the contract
changes) would make every warm follow-up relaunch Codex or OpenCode,
because the revision changes on every follow-up.
## What Changed
- `attachRun` carries `completionContract` through `NativeSession`, the
harness backend, the Codex harness session and the runnerd transport.
`executeNativeSession` passes the current input's contract. The
transport uses it instead of its first-run `resumeCompletionContract`.
- Runnerd (`codex_provider.rs`, `attach_run_in_place`): a changed
contract no longer forces a cold resume for `codex` and `opencode`.
Codex app-server never receives the contract, and the OpenCode proxy now
receives it per turn. Other providers keep the cold-resume boundary. A
changed tool set still forces it.
- Runnerd sends `paperclipCompletionContract` with each OpenCode
`turn/start`, next to the existing `paperclipRunId`.
- The OpenCode proxy passes that contract to `attachRun` when the run
changes. `OpenCodeHarnessSession.attachRun` replaces its envelope's
contract and validates it.
## Verification
- `src/live/runnerd-codex-transport-warm-attachment.test.ts` (new case):
real runnerd and fake Codex. After a warm attachment with a new
contract, the saved `runAttachTemplate` has the new revision and the
Codex pid is unchanged.
- `src/cli/opencode-proxy-input.test.ts`: the real proxy bundle serves a
warm second run with revision `proxy-v2`. The second
`paperclip/runResult` is produced and names `proxy-v2`. Without the fix
the finish call is refused.
- Each test fails without its part of the fix: the proxy test without
the proxy change; the transport test without the transport change (stale
saved revision); and, with `PAPERCLIP_ATTACH_TRANSITION_RUNNER` pointed
at a runnerd built without the `codex_provider.rs` change, the Codex pid
changes.
- `cargo test -p paperclip-runner-core --lib warm` (11 tests),
`native-session-runtime.test.ts` (232), the OpenCode driver, backend and
warm-attachment suites, and `tsc --noEmit` for the runner package pass.
## Risks
- Previous-release story (`rules/upgrade-safe-run-state.mdc`): no
persisted shape changes. Saved `runAttachTemplate` values carry forward,
and the next attachment writes the current contract. A new runnerd with
an old OpenCode proxy behaves as today: the proxy ignores the new
parameter and keeps the stale contract. An old runnerd sends no
parameter, so a new proxy keeps its launch contract.
- In-place adoption widens the warm-retention path for Codex and
OpenCode to changed contracts. The idle, accounting and pending-work
checks still apply first.
> 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
|
a9ac2e2039 |
feat(native): split local runner startup into stage, launch and session-load spans (#19852)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - When a person messages an agent, the agent must start its turn fast. Time to first token (TTFT) and agent boot are first-class metrics. > - The Phase 0 baseline (#19846) found that local runner startup costs 1.6–17 s at p50 and up to 50 s at p95. It is the largest part of TTFT. > - The run log records that startup as one span (`runner.session.bootstrap` or `runner.session.resume`). We cannot see if the time goes to staging, to the `paperclip-runnerd` process launch, or to the provider session load. > - This pull request splits local runner startup into three sub-spans in the run log. > - The benefit: the follow-up work to cut follow-up-turn TTFT can measure each step before and after. ## Linked Issues or Issue Description Refs #19846 (TTFT baseline, finding 1 and gap "No sub-spans exist inside local runner startup"). Refs #19838. **What existing behavior does this improve?** The run-log performance spans for local native runner startup. **Current behavior** The local path records one `runner.session.bootstrap` or `runner.session.resume` span that covers runtime staging, the runner process launch and the provider handshake together. The remote path already records `runner.runtime.stage` and `runner.process.launch`; the local path does not. **Proposed behavior** The local path records three sequential child spans of `runner.session.startup`: `runner.runtime.stage`, `runner.process.launch` and `provider.session.load`. **Reason and benefit** The TTFT work needs to know which step dominates before it changes the runner lifecycle. Without these spans, a change to warm runner reuse cannot show which cost it removes. ## What Changed - `RunnerdCodexTransport` has a new optional `onStartupStep` callback. `#start` and `#resume` time three milestones and call it synchronously. A failing callback never changes startup. A nested startup (exact adoption from `thread/start`) keeps the outer clock. - `runner.runtime.stage`: launch preparation before the runner process exists. - `runner.process.launch`: spawn until runnerd accepts the run (`run.prepare`, or the recovery attachment on resume). An adopted live runner records no launch. This keeps the existing meaning in `summarizeStartupReuse`: a launch span means a fresh runner. - `provider.session.load`: `session.open` (or the recovery probe) until the provider identity is confirmed. - `recordNativeRunnerStartupStep` (`native-run-trace.ts`) records each step as a `run.performance.span` row with `mode` and `target: "local"`. It uses `trace.defer`, so the run-log write never delays `turn.start`. - The native executor passes the callback only for a local target. Remote launchers keep their own spans, so there are no duplicates. - `redaction.ts`: allowlist `runner.process.launch` and `provider.session.load` as native span names. - `doc/run-log-events.md`: new section "Local runner startup sub-steps". Data path (AGENTS.md §5.7): run log only, plus Observability spans when an operator sets an OTLP endpoint. The attributes (`mode`, `target`, `outcome`) are already in the closed span-attribute allowlist. No Telemetry change. ## Verification - `pnpm --filter @paperclipai/paperclip-runner exec vitest run src/live/runnerd-codex-transport-startup-steps.test.ts`: a real runnerd and the fake Codex fixture. Bootstrap and a cold resume each report stage, launch and session load in order, with `ok` outcomes and no overlap. A throwing sink does not change startup. - `cd server && npx vitest run src/services/native-runtime/native-run-trace.test.ts src/__tests__/redaction.test.ts`: the rows have the correct parent, offsets, durations and failed level. A stalled run-log sink does not block the caller. The rows settle before `task.provider_session.measured`. The names survive redaction. - The existing `runnerd-codex-transport*` suites pass (412 tests). Under host load (load average 16–27), the first full run had 7 timing failures in suites that do not set the callback. Each failing file passed when I ran it again alone. `runnerd-codex-transport-settlement` passed 3/3 for the failing case alone. - `tsc --noEmit` passes for `packages/paperclip-runner` and `server`. ## Risks - Low risk. Without `onStartupStep`, the transport behaves as before: the clock is not created and each milestone call is an optional-chain no-op. - More run-log rows: three per local runner startup. - The step boundaries are controller-side observations. `runner.process.launch` includes the runnerd authentication and the `run.prepare` acknowledgement, not only the OS spawn. > 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
|
b5b972ceb1 |
fix(runner): stop provider tool processes that leave the owned process group (#19727)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - The local runner (`paperclip-runnerd`) supervises each provider process, such as the ACPX sidecar for Claude, in its own process group > - Run teardown stops that process group, and nothing else > - Claude Code starts each Bash tool command in a new session (`setsid`). A command such as `nohup ./bench.sh & echo $!` leaves a job in that session, and the wrapper exits at once. The job is then a child of init and outside the owned group > - In a restart chaos storm, a run failed after a restart resume. Its tool job kept running beside the task's next run, in the same workspace, for more than four minutes > - This pull request makes the supervisor record every process that its provider starts, including processes that leave the group, and stop them all before teardown reports success > - The benefit is that a run's tools cannot outlive the run and cause duplicate side effects in the next run ## Linked Issues or Issue Description **What happened?** A restart chaos storm ran a local ACPX Claude task. After a restart resume, the agent ran `nohup ./bench.sh <nonce> > log 2>&1 & echo "started pid $!"` and then ended its turn without a result. The run failed with `native_finalization_missing`. The runner tore down the sidecar group, and the guardian killed the provider group. `bench.sh` was in the session that Claude Code made for the tool (pgid and sid = the exited wrapper, ppid = 1). It was never signalled. It kept forking `sleep 1` while the task's next run started and finished in the same workspace. **Steps to reproduce** 1. Run a local ACPX Claude task whose agent runs a Bash tool command such as `nohup ./bench.sh x > log 2>&1 & echo $!`. 2. End the run (any outcome, including failure after a restart resume). 3. Run `ps -axo pid,ppid,pgid,command | grep bench.sh`. The job is still running with ppid 1. **Agent adapter(s) involved** Claude Code (local ACPX). The gap is in the shared supervisor, so any provider whose tools leave the group is affected. **Root cause** `SupervisedProcess` (`packages/paperclip-runner/runner/crates/runner-core/src/process_supervisor.rs`) spawns with `process_group(0)` and stops only that group in `terminate_group`, `terminate_group_immediately`, `wait` and `Drop`. Its own comment said that a caller that must exclude escaped or re-parented descendants needs a separate lifetime fence. No caller had one. The ACPX sidecar transport (`acpx_sidecar_transport.rs`, `shutdown` and `poison`) and the ACPX provider guardian (`installation-integrity.ts`, which kills only its own group) therefore left every `setsid` tool job behind. The failure path after the resume was not special: the same gap applies to every teardown. **Expected behavior** When a run ends for any reason, every process that its provider or tools started stops, and teardown proves that they are gone. Identity comes from PID, group, session and start time, never from a command name. ## What Changed - Add the `paperclip-process-table` crate. It reads PID, parent, process group and start time for every process (libproc on macOS, `/proc` on Linux). `runner-core` forbids unsafe code, so the two checked libproc calls live in this small crate. - Add `process_lineage.rs` to `runner-core`. A sampler thread (every 50 ms) records each descendant of the supervised process by PID and start time. It also records each group and session that a descendant leads, with the leader's start time. A process belongs to the lineage when its parent belongs, or when it is in a recorded group or session and started no earlier than its leader. Groups and sessions that the supervised process only inherited (runnerd's session) are never recorded. A recorded ID whose PID now names a different process is dropped. - `SupervisedProcess` samples the lineage just before it signals, and after the leader is reaped it stops every live member: TERM, then KILL after the shutdown grace. It fails with an error until it proves that no member is alive (grace plus 5 s). This applies to `terminate_group`, `terminate_group_immediately`, `wait` and `Drop`. - If the lineage sampler cannot start, the supervisor stops the process it just started and reports the failure, so a provider never runs without tool cleanup. - Table reads and state updates happen under one lock, so an older read cannot erase members that a newer read recorded. - The process table reports processes that exist but cannot be read. A recorded member that cannot be read stays unresolved, and teardown fails until its exit is proven. - The Runner workspace now has two members. `scripts/runner-release-crate-cache.mjs` and `scripts/runner-debug-crate-cache.mjs` cache and mark current the process-table lib unit with the Runner crate (otherwise Cargo rebuilds the Runner crate after every restore). The debug contract accepts only this library-only member and still refuses any other workspace change. The workflow cache path lists and the debug clean step are updated to match. - `terminate_group_immediately` now accepts `EPERM` from the group signal when the leader has already exited. macOS returns `EPERM` for a group whose only member is the unreaped leader. The lineage pass then proves that no other member remains. ## Verification - `cargo test -p paperclip-runner-core --lib -- process_lineage tool_job immediate_group_fence` (macOS, local). - New real-process tests in `process_supervisor.rs`: a supervised shell starts a wrapper with `setsid` (Perl `POSIX::setsid`), the wrapper runs `nohup <loop> &`, writes the job PID and exits. The test asserts that the job is outside the owned group and re-parented, then calls `terminate_group`, `terminate_group_immediately` or drops the process, and asserts that the job is gone when the call returns. - Red: with the tracker disabled, all three fail with `tool job <pid> outlived its provider`. - Green: all three pass. - Unit tests in `process_lineage.rs` cover a re-parented job, a job that left its group but stayed in the recorded session, a session made after the first sighting, PID reuse, and no claim through the supervisor's inherited session. - `node --test scripts/__tests__/runner-release-crate-cache.test.mjs scripts/__tests__/runner-debug-crate-cache.test.mjs` (new case: the process-table library is cached with the crate, and any other member still needs review). - `cargo test -p paperclip-process-table` (includes an exited process reported as gone), and `cargo test -p paperclip-runner-core --lib` (570 passed). `acpx_provider_backend::tests::unconfirmed_suspension_state_becomes_recoverable_only_after_cleanup_and_suspend` fails sometimes under full-suite load. It fails at the same rate with the tracker disabled, so it is a flake that already existed. - Sampler cost: one table read takes about 0.4 ms for 1,200 processes on macOS (about 0.8% of one core for each supervised process). - The restart-chaos replay (`--replay w-20261107T023921-s7106/t67`) did not run. The local session was not permitted to load provider credentials. It is not proven against a live Claude run. ## Risks - Teardown now stops all the processes that a provider started, also intentional background jobs. This is the intended contract: nothing outlives the run. - A wrapper that lives less than one sample (50 ms) and orphans a job before the sampler sees it is not recorded. Claude Code's wrapper sources a shell snapshot and lives for seconds. On Linux, a child-subreaper would close this gap. That is not in this change. - When `runnerd` itself is killed with SIGKILL, this code does not run. The provider guardian still kills its own group, but escaped tool sessions are not stopped in that case. - Teardown can return a new error when it cannot prove the lineage stopped. Callers already handle `terminate_group` errors as provider shutdown failures. - The server does not yet gate the next run on the tool lineage of the prior run directly. It relies on the runner finishing its teardown before the runner exits. ## 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
|
b2af45f0f1 |
fix(runner): let OpenAI API-key Codex agents generate images with their existing key (#19654)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work.
> - New users often connect a Codex agent with an OpenAI API key during
onboarding.
> - A user asked such an agent for a logo. The agent said that the
built-in image generator was not available and that its fallback needed
a separate API key. It then drew an SVG by hand.
> - #19516 removed `--disable image_generation`. That change is correct,
but it does not reach this path. The pinned Codex 0.160.0 shows its
`image_gen` tool only when Codex signs in with ChatGPT. The check is
`image_generation_available` in `codex-rs/core/src/tools/spec_plan.rs`.
An API key never gets the tool.
> - The runner keeps the provider key out of the agent's shell on
purpose, so the skill's `scripts/image_gen.py` fallback cannot run.
> - This pull request adds a Paperclip `generate_image` tool. The server
makes the image with the agent's existing OpenAI key. The key never goes
to the agent process or the shell.
> - The benefit is that API-key Codex agents can make images with no
second key. The spend goes to the agent's budget and the budget can stop
it.
## Linked Issues or Issue Description
Refs #19516
**What happened?**
An agent connected through onboarding with an OpenAI API key (Codex
harness, `gpt-5.6-sol`) could not generate images. The agent said: "The
image skill's built-in generator isn't available in this run, and its
fallback would require a separate API key." It then drew an SVG instead.
**Steps to reproduce**
1. Create a company with a Codex agent that uses an OpenAI API key.
2. Ask the agent for a raster image, for example a logo PNG.
3. The agent reports that image generation is not available. It does not
make the image.
**Expected behavior**
The agent makes the image with the OpenAI key that already powers it.
**Root cause**
The agent ran on the native runner's Codex app-server path, with
`auth.json = {"OPENAI_API_KEY": ...}`. Codex 0.160.0 removes
`image_gen.imagegen` from the model's tools unless
`current_auth_uses_codex_backend()` is true. That function is true only
for a ChatGPT sign-in.
I proved this with the real pinned binary. I gave it the runner's
isolated launch arguments and the API-key auth file, and pointed it at a
local Responses endpoint. The first model request offers only `exec`,
`wait`, `request_user_input` and the collaboration tools. #19516's test
checked only `codex features list`, so it could not see this gate.
## What Changed
- **New `server/src/services/native-runtime/generate-image-tool.ts`:**
- The tool schema and definition.
- `imageGenerationBindingForRun`. It applies only to native Codex runs
on a direct OpenAI `api_key` connection with no router selection.
- The OpenAI Images API call (`gpt-image-2`, `b64_json`, PNG check, 180
s timeout). The reply is read as a stream with a byte cap before it is
parsed. A decoded image larger than 20 MiB is refused.
- A reviewed rate card for `gpt-image-2`: text input $5, image input $8
and image output $30 per 1M tokens, from
developers.openai.com/api/docs/pricing, retrieved 2026-11-06.
- A create-only, symlink-confined local write.
- **`paperclip-runner-tool-authority.ts`:**
- Adds the `imageGeneration` binding.
- Advertises the tool only for standard mode with a workspace. A remote
target must also have an installer. Dot, Muse, Grok and native review do
not get the tool.
- Adds the `#generateImage` handler. Before any spend, it checks again:
the run and task, budget authority, `issue:mutate`, case handoff, and
the connection.
- Stores a claim-before-spend per-call receipt in
`resultJson.imageToolReceipts`. A retry or recovery replays the receipt
and does not pay again.
- Saves a `charged` receipt (usage, price, request ID and charge time)
as soon as OpenAI returns. A failed save, failed accounting write or
crash can then still settle the spend. A retry or recovery records the
cost once and never buys again.
- Records one `cost_events` row through `createCostEventInTransaction`.
The row is `metered_api` with `costStatus: "estimated"`, and the budget
is evaluated. The row is recorded even if saving fails.
- Checks the task's project budget, as well as the company and agent
budgets, before spending.
- Writes content-free `runner.image_generate` activity entries: started,
completed, failed and unconfirmed.
- **`ai-connections.ts`:** adds `selectImageGeneration`. It checks again
that the connection is an active, healthy OpenAI `api_key` connection
with no custom routing. It also checks the responsible user's
membership, the grant audience and the agent or company installation.
These are the same rules that runtime selection uses.
- **`native-session-executor.ts`:** gets the binding from the run's
admitted `contextSnapshot.aiConnection`. The value comes only from the
server, never from tool input.
- **`native-session-resume.ts`:** the native tool contract changes from
`v18` to `v19` and now lists `generate_image`. A persisted provider
thread cannot add tools when it resumes, so older threads must rotate.
- **`doc/runner-api-tools.md`:** a new section about image generation
for OpenAI API-key Codex agents.
- **#19516: kept, not reverted.** Its changes have no effect on the
API-key path. They are still correct for Codex agents that sign in with
ChatGPT, because `image_gen` works there and saves to
`CODEX_HOME/generated_images`. Those agents need the read grant that
#19516 added. On the API-key path, the new tool replaces the built-in
tool. The tool's description tells the agent to use it instead of
`image_gen` or `scripts/image_gen.py`.
## Verification
**Red on `master` (
|
||
|
DottaandClaude Opus 5.5
|
5eac099fbb |
fix(agents): batch independent follow-up questions into one card (#19634)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work.
> - Agents ask users for input with the `request_human_input` tool or
the `ask_user_questions` interaction. Paperclip shows each call as one
question card.
> - In an onboarding walk, a Codex execution agent asked five
independent questions (name, menu and prices, contact, style, output).
It posted five cards, one after the other. Each card came 20-60 s after
the previous answer.
> - The tool description said: "For one question at a time, ask only the
next unanswered question and wait for its real answer." Agents read this
as a rule. The paperclip skill showed only a one-question example and
did not tell agents to batch.
> - No server cap exists. A question set accepts up to 64 questions.
> - This pull request changes the guidance. Agents now ask independent
inputs together in one card. They use a follow-up card only when a
question depends on an earlier answer. One-at-a-time stays available
when the user asks for it.
> - The benefit is fewer waits for the user: one card and one agent turn
instead of five.
## Linked Issues or Issue Description
**What happened?**
An execution agent asked a user five independent follow-up questions in
five separate question cards. Each card appeared only after the user
answered the previous card and the agent finished another turn (20-60 s
each).
**Expected behavior**
When an agent needs several independent inputs, it asks them together in
one multi-question card. The onboarding interview already does this ("1
of 4" pager).
**Steps to reproduce**
1. Complete onboarding on the Codex API path and hand off a task such as
"Create printable cake menu and logo".
2. Watch the execution agent's first turns.
3. See five sequential single-question cards.
**Paperclip version or commit**
master at
|
||
|
DottaandClaude Opus 5.5
|
2468b3f907 |
fix(native): tell a restart continuation that earlier processes may be gone (#19804)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - After a control-plane restart loses a turn, the native runtime sends a same-conversation continuation (#19501, #19699, #19771) > - A restart can also kill shell or background processes that earlier turns started > - In three restart storms, the continuation agent could not inspect its killed `bench.sh` and set the task to blocked > - The continuation prompt told the agent not to "blindly rerun an action whose outcome is unknown", so the agent did not re-run the command > - This pull request tells the agent that earlier processes may be gone, and to check the workspace outputs and re-run a command whose expected output is missing > - The benefit is that a task can finish after a restart instead of blocking on a process that no longer exists ## Linked Issues or Issue Description Refs #19501, #19699, #19771 **What happened?** Restart chaos workload B5 (Codex and Claude) failed this way three times: storm 6b t14, s7111 t16 and storm 10 t12. The agent started `bench.sh` in the background. A restart killed it. The restart continuation agent tried to find its tool session and got "process inspection is denied". It then blocked the task. The comment read: "Blocked by the Paperclip runtime restart: the benchmark session is unavailable…". It added that it "cannot verify whether the original run is still active or recover its output, so I have not rerun it". **Expected behavior** The agent knows it is in a resumed session after a restart. It checks the workspace for the command's outputs. If the expected output is missing, it re-runs the command and finishes the task. **Root cause** `packages/paperclip-runner/src/native-session-runtime.ts`, `restartContinuation`: the prompt did not say that processes from earlier turns may be gone. It told the agent to inspect each unfinished command's "current state" and not to "blindly rerun an action whose outcome is unknown". With the process gone, the agent could not inspect it, so it blocked. ## What Changed - The restart continuation prompt now says: - This is a resumed session after a Paperclip control-plane restart. - Shell commands and background processes from earlier turns may no longer be running, and their tool sessions may be gone. Do not block only because one cannot be found or inspected. - Check the workspace for the outputs or artifacts the command should have produced. If the expected output is present, use it. If the expected output is missing, re-run the command. - Do not repeat completed work. Before repeating an action with external effects (a push, a message, a deployment), check whether it already took effect. If you cannot tell, report the blocker. - The prompt does not say that re-running is always safe. Process teardown (#19727) has not merged. - `doc/execution-semantics.md`: one additive sentence. ## Verification - `native-session-runtime.test.ts`: the three restart continuation tests (Codex, ACPX, OpenCode) now check the new guidance. They were red before the prompt change and are green after it. - Guard: "uses the admitted session cwd in the task envelope" now checks that fresh turns and ordinary resumed turns do not contain the restart guidance. - `npx vitest run src/native-session-runtime.test.ts src/drivers/codex/ src/live/runnerd-restart-continuation.test.ts` in `packages/paperclip-runner`: 603 passed. - `tsc -p tsconfig.json` and `tsc -p tsconfig.test.json` for `packages/paperclip-runner` are clean. - Not run: a live restart-chaos replay with a real model. ## Risks - Low. This changes only the prompt text of the restart continuation. Ordinary turns do not change. - An agent may now re-run a command whose earlier run had partial side effects. The prompt asks it to check external effects first, and to re-run only when the expected output is missing. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, through Claude Code with tool use (shell, file edits, 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
|
ef83d566ca |
fix(native): ask for the result once after a restart continuation ends without one (#19771)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Native runs continue a turn that a control-plane restart lost, in the same conversation (#19501, #19699) > - In restart storms 5, 6b and 8, ACPX Claude runs crossed restarts, ran the continuation, and then failed with `native_session_retry_exhausted` > - The continuation completed, but Claude ended it with "I'll wait for the notification when it completes" for a background command, and did not call `paperclip_finish` > - The next same-run attempt then sent no turn at all and failed with "session returned no semantic result" > - This pull request makes the continuation ask for the result, and asks once more in the same attempt if the continuation still ends without one > - The benefit is that a restart-interrupted task settles without a Retry, and the restarts that the run survived cannot block the request ## Linked Issues or Issue Description Refs #19501, #19699, #19720 **What happened?** Storm 8, run 84c532f7 and its Retry 0faead66 (ACPX Claude, task CHA-5). The steps are the same in both runs: 1. `resume_dead_runner`, then `session.resumed`, then `turn.failed` with `provider_turn_lost_on_restore` (recoverable). 2. The restart continuation runs. Claude starts the bench in the background and ends the turn with "I'll wait for the notification when it completes". The turn completes with no `paperclip_finish`. 3. Lifecycle: "provider turn completed without a semantic result; same-run disposition recovery persisted" (attempt 2). 4. 30 s later, attempt 3 restores the session. It sends no turn. After about 2 s it fails with `native_finalization_missing: session returned no semantic result`, and the attempt count makes it `native_session_retry_exhausted`. **Expected behavior** The continuation tells the agent to finish and report its result. If the continuation still completes without a result, Paperclip asks for the disposition one time. Restarts that the run already survived do not block that request. An agent that never reports a result still fails closed. **Root cause** - `packages/paperclip-runner/src/drivers/codex/codex-harness-session.ts` `startTurn` (lines 224 and 248): the restart continuation uses the driver's one-shot disposition marker (`dispositionOnlyRecoveryAvailable`). #19699 uses the marker on purpose, so that a crash can adopt the accepted continuation turn. The marker then stays consumed with the continuation turn id. - `packages/paperclip-runner/src/drivers/codex/codex-session-terminal.ts` (line 226): only a *lost* continuation gives the marker back. A *completed* continuation with no result does not. - `packages/paperclip-runner/src/native-session-runtime.ts` (`dispositionRecoveryWasSubmitted`, about line 2761): on the next attempt, the consumed marker looks like a disposition request that already ran. Recovery replays the continuation's terminal and starts no turn. So the disposition request never happens. - `server/src/services/native-runtime/native-session-executor.ts` (lines 9755-9761 and `nativeSessionFailureDisposition` at 7768): the `semantic_result_missing` same-run retry charges a provider attempt on the same counter as restart resumes. In this run, attempt 3 was the last allowed attempt. - The continuation prompt (`restartContinuation`) did not tell the agent to report a result, or not to stop and wait for a background notification. ## What Changed - Codex-protocol driver (used by Codex and by ACPX through runnerd): when a restart continuation completes without a semantic result, the driver gives back the one-shot disposition marker. The disposition request is not a continuation, so it consumes the marker for good. - Native session runtime: after that, the same attempt sends one disposition request (`Paperclip semantic-result recovery after a restart continuation ... invoke exactly one paperclip_finish or paperclip_block`). It does this only when the driver shows that the continuation borrowed the marker and gave it back. Other drivers keep the current behavior. - Restart continuation prompt: it now says to finish in this turn, not to end the turn to wait for a background command or a notification, and to report one `paperclip_finish` or `paperclip_block`. - `doc/execution-semantics.md`: one additive paragraph. - No server change. The disposition request runs in the same attempt, so it uses no provider attempt. This avoids `native-restart-recovery.ts` (#19747). ## Verification Red before the fix, green after: - `codex-app-server-driver.recovery.test.ts` "keeps one disposition request after a restart continuation completes without a result". Red: marker stays consumed on `turn-2`. Green: the marker is given back, one disposition turn starts, and a resultless disposition turn does not re-arm. - `native-session-runtime.test.ts` "settles a restart continuation that ended without a result through one disposition turn (Codex driver)". Real Codex-protocol driver with a fake transport. Red: only one `turn/start`, then `native_finalization_missing`. Green: the disposition turn `turn-3` reports the result and the run completes. - `native-session-runtime.test.ts` "asks once for the disposition when a restart continuation ends without a result" and "fails closed when the driver keeps the continuation's disposition marker". Also run: - `npx vitest run src/native-session-runtime.test.ts src/drivers/codex/` in `packages/paperclip-runner`: 593 passed. - `src/live/runnerd-restart-continuation.test.ts` (real runnerd and fake Codex, killed runner): 5 passed. - `tsc --noEmit` for `packages/paperclip-runner`. - `pnpm verify` exits 2 because the doc change is a global CI input. Required CI supplies the broad coverage. Not run: a live replay of storm 8 t05 with Claude. The chaos harness needs provider keys from the operator's shell. ## Risks - Low to medium. A run whose continuation ends without a result now gets one more provider turn in the same attempt. The bound is one disposition request per continuation. Each continuation still needs an admitted restart, and the provider-attempt budget bounds those. - If the controller crashes after the continuation completes but before its checkpoint, the checkpoint keeps the stricter consumed marker. That case fails closed as before. - The prompt text changes for restart continuations. It does not change a protocol schema or a profile digest. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, through Claude Code with tool use (shell, file edits, 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
|
fff70fe49e |
fix(runner): relaunch a provider whose runner died before admitting it (#19784)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Native runs execute through `paperclip-runnerd`. When the control plane restarts, recovery starts a new runnerd process, and that process restores the provider (for example the Codex app server) from durable state. > - Each provider launch writes a durable startup fence before it spawns the process. Admission clears the fence. The fence stops a second runner from launching a provider while an unadmitted one may still run. > - A restart chaos wave sent SIGTERM and then SIGKILL to the control plane in quick succession. The second kill landed after the restored Codex provider was spawned and before it was admitted. > - The fence stayed in state with phase `spawned`. Every later runner refused to restore. The recovery probe's `runner.drain` failed with "provider startup ownership remains unadmitted", and the task needed a manual Retry. > - This pull request lets a new runner retire that fence when the recorded provider process and its process group have provably exited. It keeps every other fence closed. > - The benefit is that a double restart during provider relaunch continues automatically. ## Linked Issues or Issue Description **What happened?** A Codex run was active. A graceful restart (K2) suspended it, and recovery started a new runnerd that began restoring the provider. That runnerd spawned the Codex app server (pid recorded, phase `spawned`). A hard restart (K1) 9 s later killed it before admission. At the next boot, recovery chose `resume_dead_runner` and queued its `runner.drain` probe. The new runnerd failed the probe with `command_execution_failed: provider startup ownership remains unadmitted`. The run ended `native_session_recovery_failed`, and the task needed a Retry. The kept provider state had `lifecycle: "turn_active"` and a `startupAttempt` with `phase: "spawned"`, `trigger: "restore"`, `processId` = `processGroupId`, and no recorded exit. That process no longer existed. **Steps to reproduce** 1. Start a Codex task on a local runner. 2. Restart the control plane gracefully during the turn. 3. Kill the control plane again while recovery relaunches the provider, before the provider is admitted. **Expected behavior** Recovery relaunches the provider and continues the run. **Actual behavior** Every later runner refuses to restore, and the run fails. ## What Changed - `provider_backend.rs`: after a runner loads durable provider state, `retire_orphaned_startup_attempt` clears the startup fence only when all of these hold: - the attempt is in phase `spawned` (neither admitted nor failed); - the session is still restorable (`session_open`, `turn_active` or `provider_exited`); - signal 0 returns `ESRCH` for both the recorded process and its process group. It records a `provider_startup_retired` diagnostic event with the attempt, persists the state, and lets the normal restore path relaunch the provider. - Intent-phase fences (OpenCode). Three OpenCode runs after a K6 crash loop, and other runs on master, stopped at phase `intent` with no recorded process. Under load, the runner spends seconds verifying launch artifacts after it records intent. An intent alone could not show whether a process was started. - A new intent now carries `beforeSpawn: true`. - A new `SpawnStarting` observation clears the mark immediately before every provider spawn site in `start_with_tools_observed`. That clear is fsynced, but it emits no extra diagnostic event. - An intent still marked `beforeSpawn` (in a restorable session) proves that no process was started, so the fence is retired. - An intent that reached the spawn, or one written before this mark existed, stays fenced. - A runner that predates the field rejects the state as unknown (`deny_unknown_fields`), which fails closed. - `validate()` only allows the mark on an intent or on a spawn-stage failure. - The fence stays closed for: - an intent that reached the spawn, or one without the before-spawn mark; - a failed startup (initialization or spawn); - a session that recovery closed; - a process or group that may still run; - a permission error or an invalid id. - The existing test `failed_authenticated_identity_commit_keeps_the_durable_startup_fence` now uses a live process group instead of the fixed pid 123. That pid might not exist on a CI host, so the test would no longer prove a running provider keeps the fence. ## Verification - Red before the fix: the new integration test `restart_relaunches_a_provider_whose_runner_died_before_admitting_it` (fake Codex app server; turn active; a `spawned` restore fence with an exited process group; then the recovery `runner.drain` probe) failed with `provider startup ownership remains unadmitted`, the incident error. After the fix it passes. It shows one more `thread/resume`, the retired diagnostic, and a cleared fence. - Unit tests (unix-only module): retirement for each restorable lifecycle, and for an intent still before the spawn. A `SpawnStarting` outside its intent is rejected. The fence stays closed for: - an intent that reached the spawn; - an intent without the before-spawn mark; - a real child that survives its killed leader (process-group probe, no stubs); - a failed spawn; - a failed admission; - a closed session; - a live provider; - a surviving process group. - `cargo test -p paperclip-runner-core --test codex_provider`: 119 passed. The startup event counts are unchanged, because the spawn-start mark emits no event. The existing fence tests (`failed_provider_startup_is_persistently_fenced_before_another_process_can_resume`, `durable_recovery_closes_a_provider_that_reopens_a_settled_turn`) still pass. - `cargo test -p paperclip-runner-core --lib`: 570 passed. `cargo check --workspace --tests` passes. - `cargo fmt --check` passes. ## Risks - Process ids can be reused. A reused id only makes a gone process look alive, so the fence stays closed. Retirement needs `ESRCH` for both the pid and the group. - A provider process that moved to a new process group is outside this check. Before admission the runner has sent it no turn, so it cannot repeat provider work. - Intent fences written by older runners, including the kept incident states, have no before-spawn mark. They stay fenced and still need the Retry. New launches carry the mark. - A runner-side change takes effect on local runners after the runner build. Daytona needs a new image. > 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`), used through Claude Code with tool use (shell, file edits, test execution). Default context window. Extended reasoning was available. ## 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
|
35d41f11bb |
fix(runner): keep Codex thread usage reported before a run's first turn as history (#19783)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - A new Codex run can attach to a Codex thread that earlier runs used > - In restart storm 9, a comment-wake run attached to that thread after a restart. Before its first turn, it received the thread's usage from the earlier run > - The runnerd transport gave that usage an empty turn owner. The driver rejected it as an unowned authoritative notification and failed the run > - The task then stopped with `native_event_replay_conflict` and needed the next run to finish > - This pull request keeps usage that arrives before a run's first turn as history > - The benefit is that a run is not lost to earlier work's usage, and it is not billed for that usage ## Linked Issues or Issue Description Refs #19718, #19722 **What happened?** Restart storm 9 (seed 7111), task CHA-5 (Codex). Run A succeeded. A K2 restart fired. The comment-wake run B then started on a fresh runnerd and attached to the same durable Codex session. Before run B started a turn, runnerd committed a `usage.reported` event for the thread's earlier usage. The transport rehydrated it as `thread/tokenUsage/updated` with `turnId: ""`, because no turn was bound yet. `classifyCodexNotification` classifies an empty id as `invalid_authority`. The driver failed the run with "Provider authoritative notification thread/tokenUsage/updated did not name its execution owner." Recovery then stopped with `native_event_replay_conflict`. **Expected behavior** Usage that arrives before a run's first turn does not fail the run. If the session has no usage baseline yet, that usage is history, so the run is not billed for it. **Steps to reproduce** 1. Use fake Codex with `--usage-before-first-turn`. It reports thread usage right after `thread/start`. 2. Open a driver session over the real runnerd transport. Wait until runnerd commits the usage, then start a turn. 3. Before this change, the run fails with "did not name its execution owner". ## What Changed - `runnerd-codex-transport.ts` (`rehydrateRunnerdUsageNotification`): the transport omits `turnId` when no turn is bound. It no longer sends an empty owner. - `codex-session-notifications.ts`: usage reported before the session's first turn, with no baseline yet, sets the baseline (history). Usage during or after a turn keeps its existing accounting. - Fake Codex: new `--usage-before-first-turn` flag. - New test `src/live/runnerd-codex-pre-turn-usage.test.ts` with real runnerd, the real Codex driver and fake Codex. ## Verification - Red, run twice before the fix: the test failed with `NativeProviderTerminalFailure: Provider authoritative notification thread/tokenUsage/updated did not name its execution owner.` This is the live message. - Green, run three times after the fix: the turn completes. The run's usage `runDelta` is 7/2 (thread total 12/3 minus the 5/1 reported before the turn). - `npx vitest run src/drivers/codex src/live/runnerd-codex-transport src/native-session-runtime`: 991 of 992 tests pass. One settlement test timed out under parallel load. It passes alone (19/19), and this change does not touch it. - `pnpm run typecheck` in `packages/paperclip-runner` passes. `cargo fmt` is clean. - Not run: the live restart-chaos replay. It needs provider credentials, which this session does not load. ## Risks - Low risk. The accounting change applies only when no turn is active, no turn is pending, the session has no settled turns and there is no baseline. In that state, the usage cannot belong to this run's work. - The test reproduces the failing shape (usage committed before the first turn) with a fresh session. It does not reproduce the exact restart and run-attach sequence from the storm. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, used through Claude Code with tool use (shell, file editing). Default context window and reasoning mode. ## 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
|
ee1e1a3994 |
fix(runner): adopt the turn a dying controller's turn.start already started (#19772)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - A remote Runner hosts the provider (here Claude through ACPX) and
keeps working while the control plane restarts
> - A controller can die after the Runner accepts its `turn.start` but
before the controller checkpoints the accepted turn or emits
`turn.accepted`
> - The provider keeps running that turn. On reattach, recovery saw no
checkpointed active turn, and the Runner's replayed `turn/started`
failed the session with "Provider started an unexpected turn"
> - The run then failed with `native_event_replay_conflict`, replay was
blocked, and a finished issue moved from done to blocked
> - This pull request lets recovery adopt that turn when the Runner's
authenticated command result proves it is the controller's own turn
> - The benefit is that a controller crash loop no longer fails a run
that the provider is still doing correctly, and a foreign turn still
fails closed
## Linked Issues or Issue Description
**What happened?**
In a restart chaos run (Claude through ACPX on a remote Runner, crash
loop fault), a follow-up comment woke a new run. The controller sent
`turn.start` and the crash loop began about one second later. After the
restarts, the controller reattached to the same Runner and failed with
"Provider started an unexpected turn". The run failed with
`native_event_replay_conflict` and recovery required workspace repair.
**What did you expect?**
The new controller generation adopts the turn that the old generation
started, waits for its terminal, and finishes the run.
**Evidence**
The durable controller journal for the run shows command 4, `turn.start`
with `turnId: turn_lab_18ac…`, completed with `{ providerTurnId:
turn_lab_18ac…, status: accepted }`. The Runner emitted `turn.started`
for that turn 1.4 s after the command, after the old controller died.
The replacement's `session.snapshot` returned `activeProviderTurnId:
turn_lab_18ac…`. The controller's own checkpoint had no active turn and
the run had no `turn.accepted` event.
Related: #19711, #19691, #19718
## What Changed
- `codex-app-server-driver-impl.ts` `recoverSession`: when the
checkpoint has no active turn, no semantic result, and no autonomous
goal or disposition turn applies, adopt the single in-progress provider
turn if the transport proves that this run's latest completed
`turn.start` accepted that exact turn id. Any other active turn is not
adopted and still fails closed.
- `codex-session-state.ts`: a recovered session with an adopted turn
emits the missing `turn.accepted` before any of that turn's
notifications.
- `app-server-transport.ts`: new optional transport method
`controllerAcceptedTurnId()`.
- `runnerd-codex-transport.ts`: implement it from the durable command
journal (`controllerAcceptedTurnStartId`). The latest `turn.start` must
be completed and accepted, its result must name the requested turn id,
and no run attachment may come after it.
- `DURABLE_TRANSPORT.md`: document the adoption rule.
## Verification
- Red/green with the real `paperclip-runnerd` and the scripted ACPX
Claude sidecar: `npx vitest run
src/control-plane/acpx-controller-recovery.test.ts -t RC-59`. The old
controller sends `turn.start` and stops before the provider accepts it.
The provider accepts and starts the turn while no controller is
attached. A replacement controller adopts the live Runner through the
production transport and `CodexAppServerDriver.recoverSession`. Before
the fix: `activeTurnId` is null (the session then fails on
`turn/started`). After the fix: events are `turn.accepted`,
`turn.started`, `turn.completed` for the same turn, with one
`turn.start` to the provider.
- Driver tests: `npx vitest run
src/drivers/codex/codex-app-server-driver.recovery.test.ts -t
"controller that died"`. Adoption passes. With no accepted result, or a
result for a different turn, `startTurn` still fails with "Provider
started an unexpected turn."
- Journal unit tests: `npx vitest run
src/live/runnerd-codex-transport.test.ts -t
controllerAcceptedTurnStartId` (pending, failed, rejected, mismatched
turn, later attachment, later pending start all return null).
- Regression: `src/drivers/codex` (all files),
`src/live/runnerd-restart-continuation.test.ts`,
`src/live/runnerd-codex-transport-adoption.test.ts`,
`src/live/runnerd-codex-transport-release-resume.test.ts`,
`src/control-plane/acpx-controller-recovery.test.ts` pass. `pnpm run
typecheck:typescript` passes in `packages/paperclip-runner`.
- `pnpm verify` asks for broader coverage (exit 2). Required CI supplies
it.
## Risks
- Low to medium. The change only acts when the checkpoint has no active
turn and the Runner reports exactly one active turn. Adoption needs the
Runner's authenticated, run-scoped `turn.start` result for that exact
id. Without that proof, behavior does not change.
- If a `turn.start` result is still pending when the replacement reads
the thread, recovery does not adopt and fails closed as before.
- Not proven on Daytona. I did not run the paid replay.
> 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 (Anthropic), model ID `claude-opus-5-5`, used through
Claude Code with tool use (shell, file edits, 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
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
||
|
Devin FoleyandPaperclip
|
2d4f80504e |
fix(runner): retain exact ACPX checkpoints after rejected turn starts (#19762)
Preserve the exact idle ACPX conversation when a matching turn.start response rejects work with retryable SESSION_RESUME_REQUIRED. Require verified identity, no pending work, confirmed process cleanup and the provider lifetime fence before normal drain and suspension can publish a reusable checkpoint. Unknown acceptance and arbitrary closed states remain closed. The real Runner regression completes the next turn on attempts 2 and 3, requires successful terminal status for the exact submitted turn, and checks expectedIdentity on both resume requests. Controlled wire faults reject failed, cancelled and missing-identity recovery. Validation: 193 affected Rust tests on unchanged Rust source/fixtures; 134 recovery integration tests, 14 environment-gated skips; 21 Claude effort tests; all server typecheck layers. Exact-head CI: 59 successful checks and statuses, 6 intentional skips. Greptile Apex 5/5, no unresolved findings. Hosted deployment and task acceptance remain separate. Fixes: #19751 Refs: #19366 Refs: #19745 Co-Authored-By: Paperclip <noreply@paperclip.ing> |
||
|
Nicky Leach
|
8be6d70538 | fix(runner): select Claude effort again after a cold session resume (#19708) | ||
|
Nicky Leach
|
167e39c4ef | fix(runner): keep warm run.attach events on ordinary accounting (#19707) | ||
|
Nicky LeachandClaude Opus 5.5
|
bf45f66619 |
docs(runner): separate the legacy-adapter paragraph in the Runner README (#19753)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - The native Runner package has a README that explains its contracts > - One section explains how a resumed run gets a fresh-conversation handoff > - The next text, about legacy adapters, has no blank line before it > - Markdown renders the two texts as one paragraph, so the legacy-adapter text looks like part of the native handoff text > - This pull request adds the missing blank line > - The benefit is a README that separates native and legacy behavior clearly ## Linked Issues or Issue Description **Issue type** Formatting error in package documentation. **Where is the issue?** `packages/paperclip-runner/README.md`, the paragraph that starts "Legacy adapters advertise `supportsToolRefreshOnResume`". **What's wrong?** The paragraph has no blank line before it. Markdown joins it to the previous paragraph about the native fresh-conversation handoff. **Suggested fix** Add a blank line before "Legacy adapters advertise". ## What Changed - Add one blank line in `packages/paperclip-runner/README.md` before the legacy-adapter paragraph. ## Verification - `git diff --stat`: one file, one line added. - I read the changed section. The legacy-adapter text now starts its own paragraph. - No code changed, so no tests are necessary. ## Risks - Low risk. This is a documentation-only change. - The Runner build-input digest includes every file in `packages/paperclip-runner`. This change therefore changes that digest, and the next release builds a new Runner image. The Runner behavior does not change. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`) in Claude Code, with tool use (shell, 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 - [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
|
fc7e3a6503 |
fix(recovery): never replace a native session after a deterministic refusal (#19733)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - Native agent runs keep a saved provider session, and same-run
recovery resumes it after a failure
> - When resume fails and the checkpoint allows it, recovery archives
the saved continuity state and opens a fresh replacement session
> - Recovery used that path for every resume failure, including
deterministic refusals that left the saved state unchanged and repeat on
every retry
> - In five staging tasks a refusal (a protected launch argument, or a
predecessor tool call with no result) escalated into a replacement. The
remote archive move failed part way, the next attempt found no runner
state, and each task was blocked for workspace repair
> - This pull request stops before replacement after a closed-class
refusal, keeps the saved state, and hands the run to an operator with a
named reason
> - The benefit is that a deterministic refusal can no longer damage
saved state or open a replacement over an unsettled external action
## Linked Issues or Issue Description
Refs #19659, Refs #19516, Refs #19334
**What happened?**
Five Codex tasks on Daytona sandboxes failed on staging. The first
attempt failed with a deterministic refusal:
- `run.attach cannot change protected launch arguments` (four tasks,
fixed separately in #19659), or
- `native_runner_source_semantic_reconciliation_required` with a
predecessor `paperclip_call_assigned_tool` call still pending (one
task).
The same-run retry then opened a replacement session.
`archiveContinuityState` failed with
`runner_continuity_break_archive_failed` and no detail. The third
attempt failed with `runner_remote_state_unavailable: exit=1`, and the
task was blocked with "Workspace repair required".
**Steps to reproduce**
1. Save a native session whose checkpoint policy is
`allow_replacement_after_resume_failure`.
2. Make the resume fail with a refusal that does not change state (an
attach refusal, or a pending tool call).
3. Same-run recovery opens a replacement session and archives the saved
continuity state.
**Expected behavior**
A deterministic refusal stops the run for an operator. The saved state
stays intact. A run with a pending side-effecting call never reaches
replacement. Transport drops keep the existing retry behavior.
**Root cause**
`executeNativeSession` checked only the checkpoint's replacement policy.
The closed failure class (#19334) only had an effect when replacement
was forbidden. Attach refusals and pending tool calls had no class at
all, and the source-reconciliation refusal was a plain `Error`.
## What Changed
- Runner (`provider_backend.rs`, `durable/mod.rs`): `run.attach`
refusals that would change the durable Codex profile or a protected
launch argument carry the closed classification
`provider_attach_refused`. An attach blocked by
`durable_pending_tool_calls` carries `provider_side_effect_pending`.
These refusals already leave the saved state unchanged.
- Contracts (`harness-driver.ts`): new closed classes
`PROVIDER_ATTACH_REFUSED` and `PROVIDER_SIDE_EFFECT_PENDING`. A typed
`NativeRunnerSourceReconciliationRequiredError` replaces the plain
`Error` at the five `native_runner_source_*_reconciliation_required`
throw sites. `runnerRecoveryFailureClass` reads the authenticated
command-result classification or the typed error, never message text.
- Runtime (`native-session-runtime.ts`): after either class, recovery
never opens a replacement session, whatever the checkpoint policy.
- An attach refusal ends with the existing
`NativeProviderStateIncompatibleError`, which maps to the existing
operator path `native_provider_state_incompatible`.
- A pending side effect ends with the new
`NativeProviderSideEffectPendingError`.
- Server (`native-session-executor.ts`, `native-session-resume.ts`,
shared `EXECUTION_RECONCILIATION_CAUSES`):
`native_provider_side_effect_pending` is a permanent failure and an
execution-reconciliation cause, so the board gets a reconciliation
(record whether the action ran) instead of a generic Retry. The
reconciled continuation does not rebind the saved session, which still
refuses attachment. It starts a fresh provider session in the same
workspace. The saved session and workspace are unchanged.
- Archive (`archiveContinuityState`): a failed remote move now records
the bounded (512 characters), redacted exit status and output in
`runner_continuity_break_archive_failed: exit=… timedOut=… <output>`.
The remote move already ran before the local move on `master`, so a
failed remote move leaves the local state in place.
`restart-interrupted-turn.ts` accepts the detailed message for its
reattachment-loss retry.
- Docs: `doc/execution-semantics.md` and
`packages/paperclip-runner/docs/durable-recovery.md` describe the new
rule.
## Verification
Red before the fix, green after (each test first run against unchanged
source from `origin/master`
|
||
|
DottaandClaude Opus 5.5
|
d1ee25d0f7 |
fix(runner): fail a pending Codex turn start at once with the runner's own terminal failure (#19722)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - The control plane starts each Codex turn through runnerd and waits for the turn's start event (`#startTurnWork` in `runnerd-codex-transport.ts`) > - In restart storm 5, runnerd failed an accepted continuation turn before it started (`turn_binding_mismatch`, fixed at its source in #19718) > - The transport held that terminal for the pending start and waited the full 30 s bound > - It then reported "runnerd did not report the provider turn identity", which hid the real cause > - This pull request makes a pending start fail at once with the runner's own terminal failure > - The benefit is a faster failure that names its real cause, so recovery and operators act on the correct fact ## Linked Issues or Issue Description Refs #19718, #19699, #19608 **What happened?** Restart storm 5 (seed 7106), task t01 (Codex). Runnerd accepted the restart continuation turn. About 0.5 s later it committed a `harness.diagnostic`, a non-recoverable `turn.failed` (`turn_binding_mismatch`) and a failed `run.terminal`. The transport's event pump holds a turn terminal while a turn start is pending, so the start loop never saw it. The start waited 30 s, the #19608 adoption check found no active turn, and the run failed with "runnerd did not report the provider turn identity". **Expected behavior** When runnerd reports a final failure of the pending turn (a non-recoverable `turn.failed` or a failed `run.terminal`) before the turn starts, the start fails at once. The error names the runner's code and message. **Steps to reproduce** 1. Use fake Codex with `--foreign-usage-before-turn-started --turn-started-delay-ms 20000`. 2. Start a turn through the runnerd Codex transport. 3. Before this change, `turn/start` rejects after the full bound with "runnerd did not report the provider turn identity". ## What Changed - `runnerd-codex-transport.ts`: new `runnerdPendingTurnStartFailure()`. It finds a non-recoverable `turn.failed` or a failed `run.terminal` among the terminals held for the pending start and the committed events after the cursor. - `#startTurnWork` checks it while it waits for `turn/started`, before the adoption snapshot, and in the post-adoption wait. It throws `runnerd failed the provider turn before it started: <code>: <message>`. - The check runs only when the turn has not started. A recoverable terminal does not trigger it. The #19608 late-start adoption path is unchanged. - Fake Codex: new `--foreign-usage-before-turn-started` flag. It sends usage for a turn the runner does not own right after it accepts a turn. This makes runnerd fail the turn with `turn_binding_mismatch` before it starts. - New test in `runnerd-codex-transport-deadlines.test.ts` with real runnerd and fake Codex. ## Verification - Red before the fix: `npx vitest run src/live/runnerd-codex-transport-deadlines.test.ts` → the new test failed after 16.4 s with "runnerd did not report the provider turn identity". - Green after the fix: the start rejects in under 5 s with `turn_binding_mismatch`, and fake Codex saw exactly one `turn/start`. - The #19608 tests ("adopts a durably accepted turn whose start notification outlives the turn-start bound" and "does not adopt an accepted turn the runner no longer reports active") still pass. - `npx vitest run src/live/runnerd-codex-transport`: 9 files, 399 tests pass. - `pnpm run typecheck` in `packages/paperclip-runner` passes. `cargo test --bin fake-codex-app-server` passes. `cargo fmt` is clean. - `pnpm verify` exits 2 (broad input: the fake Codex binary). Required CI supplies the broad coverage. ## Risks - Low risk. The new failure fires only when runnerd itself already reported a final failure and no turn start was observed. In that state the start could not succeed, and before this change it failed after the bound anyway. - If a provider sends a non-recoverable terminal before its own turn start for a turn that later starts, the start now fails earlier. Codex sends `turn/started` before its terminal, and runnerd's terminal for a protocol failure is final, so I do not expect this. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, used through Claude Code with tool use (shell, file editing). Default context window and reasoning mode. ## 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
|
42f6b3a870 |
fix(runner): keep a restart-lost Codex turn's late usage from failing its continuation (#19718)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Native Codex runs continue a turn lost to a control-plane restart once, in the same conversation (#19699) > - In restart storm 5, a Codex run crossed one graceful restart. The continuation started, but the task stopped and needed a Retry > - The runner log shows the cause. The resumed Codex thread sent the usage of the lost turn after it accepted the continuation turn. The runner rejected that usage as `turn_binding_mismatch` and failed the run > - Recovery marks the lost turn as settled in the durable ledger. It does not tell the live Codex provider, so the provider treats the old turn's usage as a foreign event > - This pull request records the lost turn as settled in the live provider too > - The benefit is that a restart-lost Codex turn continues without a Retry ## Linked Issues or Issue Description Refs #19699, #19608, #19501, #15297 **What happened?** Restart storm 5 (seed 7106). Task t01 (Codex, B5) crossed one K2 SIGTERM restart. Recovery resumed the dead runner. Runnerd restored the conversation and recorded `turn.failed` with `provider_turn_lost_on_restore` (recoverable). The control plane started the continuation (#19699). Runnerd accepted it as a new provider turn. About 0.5 s later, Codex sent `thread/tokenUsage/updated` for the lost turn id. The runner's durable state records these events: - `turn.accepted` for the new turn `01a1b3e2-…` - `harness.diagnostic` `turn_binding_mismatch`, method `thread/tokenUsage/updated`, `expectedTurnId` = new turn, `receivedTurnId` = lost turn `01a1b3e0-…` - `turn.failed` (`recoverable: false`) and `run.terminal` failed The control plane did not see `turn/started` and failed after 30 s with "runnerd did not report the provider turn identity". The task needed a manual Retry. **Expected behavior** Usage for the lost turn is history. The runner keeps it as a resume usage snapshot. The continuation runs, and the task continues without a Retry. **Steps to reproduce** 1. Start a Codex turn through the durable runner. 2. Kill the runner while the turn is active. Restore so that the provider reports no active turn (`provider_turn_lost_on_restore`). 3. Start the continuation turn. Let the provider send `thread/tokenUsage/updated` for the lost turn after it accepts the new turn. 4. Before this change, the runner fails the continuation with `turn_binding_mismatch`. ## What Changed - `provider_backend.rs`: when recovery finds that the previous active turn ended while the runner was offline, it now also restores that turn id into the live provider's settled-turn set. The durable ledger already settled it. The existing settled-turn path then turns its late usage into `paperclip/resumeUsageSnapshot` and does not bill its `last` usage to the run. - The code skips this step when the settled-turn epoch needs a rollover, the same as the durable ledger. - Fake Codex: new `--stale-usage-after-turn-start` flag. With durable turn ids, it sends usage for the previous turn right after it accepts the next turn, before `turn/started`. This matches the live Codex order. - New test `restart_continuation_ignores_late_usage_for_the_turn_lost_on_restore` in `tests/codex_provider.rs`. It uses the real durable Codex executor with fake Codex. ## Verification - Red before the fix: `cargo test -p paperclip-runner-core --test codex_provider restart_continuation_ignores_late_usage` failed with the same diagnostic as the live run (`turn_binding_mismatch`, method `thread/tokenUsage/updated`, `receivedTurnId: provider-turn-1`, `expectedTurnId: provider-turn-2`). - Green after the fix: the same test passes. The continuation emits `turn.started` for `provider-turn-2`, there is no `turn.failed` or `run.terminal`, and fake Codex saw exactly two `turn/start` calls (the original and one continuation). - `cargo test -p paperclip-runner-core`: all suites pass. `acpx_park_readiness::hosted_park_acpx_preserves_late_goals_and_rejects_foreign_or_busy_observations` failed once and passed on rerun. This change does not touch ACPX. - `cargo fmt --check` is clean. - Not run: the live restart-chaos replay of `w-20261107T023921-s7106/t01`. The local harness needs provider credentials that were not available to this session. ## Risks - Low risk. The change adds one already-ended turn id to the live provider's settled set. That set is only used to ignore stale events and to reject reuse of a settled id. So a provider that later reports the lost turn id as a new active turn is now rejected, as the durable ledger already does. - The change does not address a second, separate problem: after the runner failed the continuation, the control-plane transport waited the full 30 s turn-start bound and reported "runnerd did not report the provider turn identity" instead of the runner's `turn_binding_mismatch`. That message only applies when such a failure occurs. This PR removes this specific trigger. ## Model Used - Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, used through Claude Code with tool use (shell, file editing). Default context window and reasoning mode. ## 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
|
ece09c156a |
fix(runner): keep an accepted result when ACPX reports its turn lost on restore (#19691)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Native Codex, OpenCode, and ACPX runs survive control-plane restarts: runnerd restores the conversation, and the control plane continues the task from its checkpoint > - #19658 made recovery keep a durable result when a Codex restore loses the turn that proposed it. It covered the recovery read path only > - ACPX reports the lost turn differently. After the restored conversation reopens, a live `turn.failed` arrives with `provider_turn_lost_on_restore` > - That live terminal still turned an accepted `done` result into a failed run. The run ended `failed` (`adapter_failed`), and the finished work needed a Retry > - This pull request moves the rule into the shared terminal mapping. A restore-lost turn with a held result for that exact turn settles as completed, whichever path reports it > - The benefit is that ACPX and OpenCode tasks that already finished keep their accepted result after a restart, as Codex tasks now do ## Linked Issues or Issue Description Refs #19658, #19525, #19501, #19539 **What happened?** In a restart chaos wave, an ACPX Claude task (workload A1) wrote its brief and proposed a `done` result (`run.result.proposed`, then a completed `paperclip_finish` tool call). Then a K1 and a K6 crash loop hit the control plane, and `resume_dead_runner` ran at controller generations 2 and 3. The restored ACPX conversation emitted `turn.failed` with `provider_turn_lost_on_restore` (`recoverable: true`) for the checkpointed turn. The control plane then emitted `run.result.accepted` with the `done` result, but `run.terminal` came back `runTerminalState: failed` / `turnTerminalState: failed` / `reportedWorkDisposition: done`. The run ended `failed` / `adapter_failed`, "Native session failed". **Steps to reproduce** 1. Start a native ACPX (or OpenCode) task through runnerd. 2. Let the agent call `paperclip_finish`, and let the control plane persist the proposal. 3. Kill the control plane and runner before the turn ends. 4. Let restart recovery resume the run. The restored runner reports the turn lost on restore as a live terminal. **Expected behavior** Recovery finalizes the accepted result, and the run succeeds. **Actual behavior** ACPX and OpenCode use the same Codex harness session as Codex (`codex-native-backend.ts`). Recovery restored the held result for the turn from the durable proposal. But the live `turn/completed` notification with `status: failed` and `provider_turn_lost_on_restore` went through `mapTerminalTurn`, which emitted `turn.failed`. The backend maps `turn.failed` to `runTerminalState: failed`. #19658 handled only the `thread/read` reconciliation path. ## What Changed - `codex-session-terminal.ts`: a new `settleRestoreLostTurnWithHeldResult` helper. When a terminal is `failed` with `provider_turn_lost_on_restore` and the session holds the semantic result that this exact turn proposed, the turn settles as `completed`. `mapTerminalTurn` applies it first, so live notifications and recovery reads follow one rule. - `codex-session-terminal.ts`: the helper also covers `acpx_active_turn_recovery_closed`. Runnerd's ACPX backend (`acpx_provider_backend.rs`, `restore_session_if_needed`) emits that code instead of restoring the conversation when its own state already recorded the accepted result. It is the same restart loss, with the result already accepted. #19525 already keeps accepted results when close cannot prove suspension, and session close still handles `provider_exit_unconfirmed`. - `codex-harness-session.ts`: `reconcile()` keeps only the missing-turn case from #19658. The restore-lost read case and the prior-terminal check now use the shared helper. - `doc/execution-semantics.md`: the restart paragraph now states that the rule covers Codex, OpenCode, and ACPX, and both report paths. ## Verification - Red before the fix, with the real Codex native backend (`createCodexNativeSessionBackend`) and a fake transport through `executeNativeSession`: `pnpm --filter @paperclipai/paperclip-runner exec vitest run src/native-session-runtime.test.ts -t "journaled completion proposal"`. The new `restore-lost live` scenario failed. Green after the fix: the run completes with the journaled result on `turn-1` and a succeeded terminal, and no `turn/start` is sent. The `restore-lost read` scenario passes on both (#19658 already covered it). - Driver red/green: `src/drivers/codex/codex-app-server-driver.recovery.test.ts -t "live restore-lost"`. With a held result, the live lost terminal was `turn.failed` before the fix and is `turn.completed` after it. With no held result, it stays `turn.failed` with the lost cause, so the #19501 continuation is unchanged. - The driver case was extended to `acpx_active_turn_recovery_closed` (red with a held result before the extension, then green). It was found in a live replay where the run ended `failed`/`adapter_failed` with an accepted `done` result. - Suites: `src/drivers/codex/` and `src/native-session-runtime.test.ts` (580 tests), plus `src/backends/` and `src/live/runnerd-restart-continuation.test.ts`, pass. `typecheck:typescript` passes. - Live chaos replays (ACPX Claude): results and a separate new signature are in a comment below. ## Risks - The rule needs a held result whose turn id equals the lost turn. Recovery restores that result only from the durable `run.result.proposed` for the checkpointed turn, so a result for another turn does not settle it. - A restore-lost turn without a held result keeps its failure and the bounded same-conversation continuation. - Real provider failures and other terminal states are unchanged. - No schema, profile, pack contract, or database changes. ## 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> |
||
|
Nicky Leach
|
03498f90ee | Add explicit Runner session release and resume support (#19683) | ||
|
DottaandClaude Opus 5.5
|
a894b11a2b |
fix(runner): retry a rejected runner suspension and log every cleanup quarantine (#19653)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - A native run's close must prove that its runner durably suspended
before the session's next run may start. If it cannot, the runtime
quarantines the session (`NativeSessionCloseUnrecoverableError`)
> - In a restart chaos storm, an ACPX Claude run succeeded and its
normal close quarantined its own session. The follow-up run on that
session failed 2 s later with `native_session_cleanup_quarantined`, and
no control-plane restart came between the two runs
> - The close sent `runner.suspend` once, the ACPX sidecar rejected it
right after the turn completed, and the barrier gave up at the first
rejected command. No log line showed the quarantine or its cause
> - This pull request retries a rejected suspension within the close
deadline, and logs every cleanup quarantine with its cause when it is
created
> - The benefit is that a successful run no longer blocks its own
follow-up because of a one-time suspension rejection, and any remaining
quarantine can be traced
## Linked Issues or Issue Description
**What happened?**
Run A (ACPX Claude) succeeded. The run log of its close shows:
- `runner did not prove durable suspension before checkpoint`
- `native_session_settlement_incomplete
{"runnerSuspended":false,"providerDrained":true,"suspensionState":{"commandStatus":"failed","runnerLifecycle":"suspended",...},"semanticTools":{"pending":[]},"finalProviderState":{"activeProviderTurnId":null,"providerSettled":true}}`
The runner journal shows `runner.drain` completed, then `runner.suspend`
failed with `failed to suspend ACPX provider: ACPX sidecar command
session.suspend was rejected (retryable=false,
classification=unclassified)`. The follow-up run B on the same session
was created 200 ms after A finished and failed 2 s later at admission
(`retryQuarantinedSessionCleanups`, operator recovery branch). The
server log had no line for the quarantine. One Retry completed the task.
**Steps to reproduce**
1. Run an ACPX task whose sidecar rejects the first `session.suspend`
after the turn completes.
2. Start a follow-up run on the same session.
**Expected behavior**
The close asks again, proves suspension, and the follow-up runs. If a
quarantine is still needed, it is logged with its cause.
**Actual behavior**
The close fails on the first rejection, quarantines the session, and the
follow-up fails with no logged cause.
## What Changed
- `runner/.../durable/runner.rs` (`process_command`): a failed
`runner.suspend`, on first execution or replay, now keeps the runner in
`Continue` instead of `Suspend`. Before, the runner recorded `suspended`
over a provider whose session was still open, then exited. The RC-34
journal shows exactly that: runner lifecycle `suspended`, ACPX provider
`session_open`. A failed `runner.shutdown` still stops the runner.
- `durable-prp-control-plane.ts`: a failed or rejected `runner.suspend`
now lifts the connection's terminal-lifecycle fence, so the next pending
command (the fresh suspend) is delivered on the same connection.
Completed suspensions and every shutdown keep the fence. New test:
"delivers a fresh suspension after a rejected one on the same
connection". Before the fix, the retry was never sent and the test timed
out waiting for the frame.
- `runnerd-codex-transport.ts` (`awaitRunnerSuspensionBarrier`): after a
rejected `runner.suspend`, the barrier sends a fresh suspend command, up
to 3 attempts, 250 ms apart, inside the same close deadline. An exited
runner or a spent budget still fails closed. A rejection changes no
state: the runner returns before it records suspension.
- `native-session-runtime.ts`: new
`onNativeSessionCleanupQuarantined(listener)`. It is called when a
cleanup quarantine is created or escalates to operator recovery, with
company, run, issue, session, `operatorRecoveryRequired`, a bounded
cause (error name, code, message) and bounded settlement facts.
- `server`: `logNativeCleanupQuarantines` logs each quarantine as a
warning. The subscription is process-wide and installed once, so closes
that fail during shutdown, after HTTP closes, are still logged.
## Verification
- Red before the change: `pnpm --filter @paperclipai/paperclip-runner
exec vitest run src/live/runnerd-codex-transport.test.ts -t "rejected
suspension"`.
- "retries a rejected suspension…" got `false`.
- "stops retrying…" saw 1 attempt instead of 3.
- Green after the change: both pass.
- Rust: new test
`rejected_suspend_keeps_the_runner_serving_its_controller` covers a
failed suspend, its replay, and an accepted retry with a fresh command.
`failed_lifecycle_commands_replay_their_terminal_transition` now expects
`Continue` for suspend and `Shutdown` for shutdown. The `pnpm run
check:runner` unit suites pass (559). `acpx_park_readiness` has two
timing tests (`< 500ms`) that fail at this host load on unchanged master
as well.
- New runtime test: "announces every cleanup quarantine with its cause
when it is created" (operator-required quarantine, cause and settlement
reported once). `src/native-session-runtime.test.ts` passes (212).
- The suspension cases in `runnerd-codex-transport.test.ts` pass, except
"proves local suspension after an event backlog… (1024 suffix deltas)".
That case also times out on unchanged master at this host load (27–35);
I confirmed this by stashing the change and rerunning.
- Server `tsc --noEmit` passes after the runner package build.
## Risks
- A provider that keeps rejecting suspension now gets three tries
instead of one, within the existing close deadline. The final outcome
and quarantine are unchanged.
- The cause of the ACPX sidecar's first rejection is not proven. The
logged evidence fits the sidecar's "not at a safe suspension point"
guard (`turnId`, tools or inputs still set) racing the just-completed
turn. This change makes close tolerate it. It does not change the
sidecar.
- Log volume: one warning per quarantine.
- The Rust change ships in the runner binary. Hosted images need a
rebuild to pick it up.
- After a failed suspend, the runner stays alive with its provider. If
every retry fails, close still treats suspension as unproven and kills
the owned runner process, exactly as before.
## 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
Refs #19532, #19609.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
||
|
DottaandClaude Opus 5.5
|
c37bc749ac |
fix(native): keep restart continuations alive through early results and repeat restarts (#19699)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Native runs continue a turn lost to a control-plane restart once, in the same conversation (#15297, #19501) > - In restart storm 3, two Codex runs each crossed one restart and still needed a Retry > - The continuation was attempted, but the driver failed it with `turn_binding_mismatch`, and the run then reported "runnerd exited before provider turn startup" > - A second restart that loses the continuation turn itself also needed a Retry, because the one-shot marker was already used > - A restart should not cost a click when continuing is safe, and a failure should name its real cause > - This pull request fixes the start race, reports the driver's own failure, and gives each admitted restart its own continuation > - The benefit is that interrupted tasks finish without a Retry, within the existing recovery budget ## Linked Issues or Issue Description Refs #19501, #15297, #19658, #19539 **What happened?** Restart storm 3 (seed 7103, RC06 mask on, load 9–12). Codex B5 Retry runs crossed one K1 restart mid-turn. After the restart, recovery resumed the dead runner. The runner restored the conversation and recorded `turn.failed` with `provider_turn_lost_on_restore` (`recoverable: true`). The control plane started the same-conversation continuation (`resumeInterruptedTurn` → `startTurn`). About 1 s later the transport closed with `protocol_failure:turn_binding_mismatch`. The run failed as `native_runner_process_exited: runnerd exited before provider turn startup`. The task needed another Retry. I reproduced it with the focused cell (`--workloads B5 --faults K1 --triggers retry`) and a temporary diagnostic. The failure is "Provider semantic result did not name the active turn": a `paperclip/runResult` reached the Codex driver while the continuation's `turn/start` was still pending and no turn was bound. The driver waits for a pending start before it handles `turn/completed`, but not before it handles `paperclip/runResult`. **Expected behavior** After a restart that loses a turn, the run continues automatically when that is safe. Each admitted restart attempt can continue once, within the provider-attempt budget. A failure names its real cause. **Steps to reproduce** 1. Run `node cli/node_modules/tsx/dist/cli.mjs tests/restart-chaos/main.ts --env local --adapters codex --workloads B5 --faults K1 --triggers retry --tasks 2 --concurrency 2 --seed 4040 --rc06-masked`. 2. When a lost turn's continuation receives a semantic result before its start settles, the run fails with `turn_binding_mismatch` and the task needs a Retry. The race is intermittent: 1 of about 14 lost turns in my runs. **Root cause** - `packages/paperclip-runner/src/drivers/codex/codex-session-notifications.ts`: the `paperclip/runResult` branch checked the active-turn binding without waiting for `state.turnStartSettled`. The `turn/completed` branch already waits. - `codex-harness-session.ts` `startTurn`: the driver's own protocol failure closed the transport. The pending start then reported the runner exit, which hid the cause. - `codex-session-terminal.ts` `mapTerminalTurn`: a consumed one-shot marker blocked every later continuation in the run, even when the newly lost turn was that continuation itself. The one-continuation allowance is per run, not per issue or session lineage: attaching a new run resets the marker. The storm runs did attempt their continuation; the start race made it fail. ## What Changed - Codex driver: a `paperclip/runResult` waits for a pending turn start to settle before it is bound, like a terminal. - Codex driver: when this session's own protocol failure ends a pending turn start, `startTurn` throws `codex_turn_start_protocol_failure: <code>: <message>` instead of the runner exit. CodexRpc errors and protocol-integrity errors keep their own types. The server now classifies this as `native_session_interrupted` instead of `native_runner_process_exited`; both are retryable. - A later restart that loses the continuation turn itself re-arms the one-shot marker: the driver resets it as it was before the first continuation. It does this when it maps the lost terminal, including during recovery from a checkpoint taken while the continuation ran. The checkpoint gate `nativeRestartInterruptedTurnId` stays strict: a consumed marker still means the allowance is used. Each continuation still needs an admitted dead-runner restart. The existing provider-attempt budget bounds the repeats, and an exhausted budget still blocks with Retry. - `doc/execution-semantics.md`: an additive paragraph. ## Verification Red before the fix, green after: - `codex-app-server-driver.events.test.ts` "binds a semantic result that arrives before its pending turn start settles". Red: `session.failed`. Green: the result binds to `turn-1` and the turn completes. - `codex-app-server-driver.events.test.ts` "names its own protocol failure when that failure ends a pending turn start". Red: "runnerd exited before provider turn startup". Green: `codex_turn_start_protocol_failure: turn_binding_mismatch`. - `codex-app-server-driver.recovery.test.ts` "offers another restart continuation when a later restart loses the continuation turn itself". Red: `null`. Green: `turn-2`. A checkpoint with a consumed marker still returns `null`. - `codex-app-server-driver.recovery.test.ts` "re-arms the continuation when recovery finds the checkpointed continuation turn lost". This recovers from a checkpoint taken during the continuation, then starts the next continuation. Red without the re-arm (the marker stays consumed); green with it. - `native-runner-restart-recovery.integration.test.ts`: the same checkpoint (continuation active, consumed marker naming it) is claimed as `resume_dead_runner`. This is regression coverage. Suites run locally on the final code: - Runner: Codex driver, `native-session-runtime`, `runnerd-restart-continuation` and contracts (745 tests). - Server: `native-runner-restart-recovery.integration.test.ts`. - `tsc` for the runner tests and the server. - A few real-process tests timed out at about 6 s while six live agents ran. Each passed when rerun alone. The settlement suite passed both with and without this change at the same load. Live (local, real Codex, macOS arm64), focused restart-chaos cell `--adapters codex --workloads B5 --faults K1 --triggers retry,time --tasks 6 --seed 4042 --rc06-masked`: 6/6 crossed a restart and passed. R1 passed for all 6, so each continued without a Retry. ## Risks - A run can now continue more than once in the same conversation, once per admitted restart. Each continuation asks the agent to reconcile unfinished actions before repeating them. The provider-attempt budget still caps the repeats. - A semantic result that arrives during a pending start now waits for that start. If the start fails, the result is still rejected by the existing binding check. - The start race was intermittent (1 of about 14 lost turns), so the live cell shows absence of the failure, not a forced repro. The driver regression reproduces it deterministically. - No schema, API or migration change. ## Model Used - Anthropic Claude Opus 5.5 (`claude-opus-5-5`) in Claude Code, with tool use, code 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
|
9a52a2dfeb |
fix(native): retry ACPX session.open startup timeouts and give cold Claude a 60 s bound (#19661)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - Native runs start an ACPX provider through Runner. `session.open` opens the provider session before any prompt. > - On a loaded host, a cold ACPX Claude `session.open` took longer than its 30-second bound. The controller reported `runner_local_connect_failed: PRP command session.open timed out`. > - The server classified this as the generic `native_session_interrupted` with `ambiguous_state`, so the task was blocked at once. A person had to click Retry, sometimes several times. In one restart-chaos storm, one task needed three Retry clicks, and on another task both Retry runs timed out too. > - The server already has a bounded same-run bootstrap retry for failures that come before provider work (#18854). It did not recognize this timeout. The 30-second bound also cut off normal slow startups. > - This pull request gives cold ACPX Claude a 60-second `session.open` bound, like Pi and OpenCode (#19487). It also retries a startup `session.open` timeout automatically on the same run, from clean state. > - The benefit is that a slow cold start on a busy host completes, or retries itself, instead of waiting for a person. ## Linked Issues or Issue Description Refs #18854 (bounded bootstrap retry for definite `session.open` failures), #19487 (60-second cold OpenCode admission), #19665 (the bootstrap proof that this change reuses). **What happened?** ACPX Claude runs failed with `runner_local_connect_failed: PRP command session.open timed out` about 35 to 40 seconds after they started. The coordinator recorded `native_session_interrupted` with `recoveryMode: "ambiguous_state"` and `providerSessionEstablished: false`. The task was blocked with a board-owned action. No provider event and no checkpoint existed. The host 1-minute load was 17 to 44, sometimes with no restart nearby. **Expected behavior** A normal slow cold start completes. A `session.open` that still times out during runner startup, with no provider event, no checkpoint and no turn, retries on the same run with backoff. At most two automatic attempts follow the first. A run where provider work may have started is never retried automatically. **Steps to reproduce** 1. Run ACPX Claude tasks on a host with a 1-minute load of about 25 or more. 2. Observe a run fail after about 35 to 40 seconds with `PRP command session.open timed out`. 3. Observe the task blocked with `ambiguous_state` and a manual Retry. **Paperclip version or commit** `master` at `6c7cc2c5d6`. ## What Changed The bound (evidence below): - `coldAdmissionTimeoutMs` gives cold ACPX Claude 60 seconds (live adoption keeps 30 seconds). The runner's Rust sidecar `session_open_timeout` and the sidecar's provider handshake (`acpxSessionHandshakeTimeoutMs`) give Claude and Pi 60 seconds too. All three bounds must agree, or the shortest one fails first. Claude's warm run attachment uses the same cold budget, like Pi. The automatic retry: - `nativeSessionFailureSourceCode` returns `provider_session_open_timeout` only for exactly `<startup failure code>: PRP command session.open timed out`. The startup failure codes are `runner_local_connect_failed`, `runner_direct_wss_failed` and `runner_ingress_unavailable`. The transport uses them only before startup completes. After startup the code is `provider_transport_failed`, which is unchanged. - `nativeProviderRecoveryEvidence` treats the cause as pre-session. Durable evidence still decides: a bootstrap retry needs no provider event (`session.started`, `turn.started` and the others) and no checkpoint. The existing budget applies: three attempts with backoff of about 30 seconds, then about 60 seconds. When the budget is used, the task is blocked with the existing "Provider startup failed on 3 attempts before any provider work began" action. - A timeout is indeterminate: the runner can open a provider session after the controller stops waiting. Live runs showed both leftovers, so the retry now starts from clean state: - `prepareNativeSessionOpenTimeoutRetry` (server) runs when the failure settles. It quarantines a local runner root only when the restart-recovery proof (`runnerdStateProvesIncompleteBootstrap`, #19665) shows that no turn began and every recorded process is gone. Any other root, and any remote runner, stays `ambiguous_state` as before. - `resetAbandonedBootstrapRecord` (runner ACPX adapter) runs on a fresh open (no runner-recorded identity). If ACPX has a saved record for the session key with no messages, no prompt and no events, it is marked `reset_on_next_ensure`. Claude saves nothing before the first prompt, so reusing such a record fails the first turn with `session_resume_required`. A record that holds any conversation is unchanged. - A chat retry of an exhausted `provider_session_open_timeout` run is accepted like the other pre-provider causes. - `doc/execution-semantics.md` and the Runner README describe the cause and the bound. ## Verification **Measured startup distribution** (ACPX Claude `runner.session.bootstrap` / `runner.session.startup` spans from restart-chaos run events, 1-minute load 8 to 44): | Source | Bound | Successful cold bootstraps (s) | Failures | | --- | --- | --- | --- | | Earlier storm evidence | 30 s | 2.9, 4.7, 7.4, 8.6, 8.7, 26.2, 27.4, 27.6, 29.2 | startups ended at 34.9 to 40.3 s (the 30 s `session.open` bound plus spawn and `run.prepare`) | | This branch, wave s3601 | 60 s | 6.3, 7.8, 8.1, 13.4, 14.4, 22.3, 22.8, 22.9, **37.3** | 0 `session.open` timeouts | | This branch, wave s3602 | 60 s | 13 samples, p50 10.5, p90 17.9, max 25.7 | 0 `session.open` timeouts | | This branch, wave s3610 (final head, load 28 to 42) | 60 s | 10.4, 11.3, 11.7, 12.0, 13.3, 14.2, 29.6, **30.4**, **30.9** | 0 `session.open` timeouts | Under load, successes clustered just under 30 s and failures sat just above it, so the 30 s bound was cutting the distribution off. With 60 s, cold starts of 30.4, 30.9 and 37.3 s succeeded that the old bound would have failed. 60 s is about 1.6 times the largest observed success, and it matches Pi and OpenCode. **Live restart-chaos waves** (local, ACPX Claude only, 4 in flight, `--rc06-masked`, real provider, K1/K2 restarts): - s3601 (8 tasks, load 19 to 32): 7/8 PASS. No `session.open` timeout. The one failure is an R3 Retry signature after K1 (`native_restart_recovery_blocked`), not this cause. - s3602 (8 tasks, load 17 to 39): 8/8 PASS. No `session.open` timeout. Three Retry clicks were for `native_provider_terminal_failed` (`provider_turn_lost_on_restore`), a different restart signature. - Forced timeout, s3609 (4 tasks, B5, K2): a local, uncommitted fault injection gave every other cold startup a 1.5 s `session.open` deadline. Result 4/4 PASS. Run 4fe6537a timed out, retried itself on the same run, and then succeeded through four K2 restarts with no person involved. Run 02799a85 hit the injection three times in a row; the budget stopped it after attempt 3, and one Retry completed it. - Before the two clean-state fixes, the same forced wave (s3607 and s3608) showed the retry reusing leftover state: `turn.start` failed with `session_resume_required`, and then "ACPX session is not open". Those runs are the red evidence for the two clean-state fixes. - Final head, s3610 (8 tasks, K1/K2, no injection, load 28 to 42): **8/8 PASS**, 0 `session.open` timeouts. The one Retry click (t06) was for `provider_turn_lost_on_restore`, a different restart signature. **Tests:** - Red before the fix: `native-startup-retry.integration.test.ts -t timeout` (embedded PostgreSQL, real executor, real resume claim). The 3 retry cases failed with `native_session_interrupted` / `ambiguous_state`. - Green: `native-startup-retry.integration.test.ts` passes 36/36. New cases: - for each startup failure code, a timeout retries on the same run with backoff, keeps the task `in_progress` with an agent-owned `resume_native_run` wake, and blocks after attempt 3 with a board-owned action; - no automatic retry when `session.started` or `turn.started` is durable, when the code is `provider_transport_failed`, or when another command timed out; - a real runner root with an opened session and no turn is quarantined; a root with a turn command stays `ambiguous_state` and intact; a remote runner is never prepared. - `codex-runtime-adapter.test.ts`: 113 pass. New cases: a 45 s delayed handshake is admitted for Claude and rejected for Codex; a never-prompted record is reset only on a fresh open; records with messages, a prompt or events are kept. - `abandoned-bootstrap-record.test.ts` (real `createAcpRuntime` with `createRuntimeStore` and a controlled ACP peer that, like Claude, cannot load a never-prompted session): without the reset, a fresh open reuses the record and the first turn fails with SESSION_RESUME_REQUIRED after `session/load`. With the reset, the provider receives `session/new` and the turn completes. - `vitest run src/drivers/acpx src/cli` in the runner: 107 files, 1,935 tests pass. - `native-session-executor.test.ts -t "session.open|startup|bootstrap"`: 144 pass. `runnerd-codex-transport.test.ts -t "cold process admission"`: 4 pass. `cargo test -p paperclip-runner-core --lib admission_budget` passes. - `npx tsc --noEmit -p server` and the runner `tsc -p tsconfig.test.json` pass. ## Risks - **This changes Runner Rust code. The Daytona runner image must be rebuilt after merge** for remote sandboxes to get the 60-second sidecar bound. Until then, a remote Claude `session.open` still fails at 30 seconds in the sidecar (as a definite failure, which already retries). A remote timeout stays operator-owned, because a remote root cannot be proven clean here. - A cold ACPX Claude start that hangs now fails after about 60 seconds instead of 30 seconds. - A timed-out `session.open` may have created an empty provider session. No prompt was sent. The retry quarantines that runner root and asks ACPX for a new session. - Same-run retries of a fresh-attempt Retry run can still fail with `isolated_task_directory_binding_mismatch`. This is RC-35 (workspace re-preparation), which has its own fix in progress. - No schema or API change. The new error code appears on runs and in run error codes. ## Model Used Claude Opus 5.5 (`claude-opus-5-5`), Anthropic, through 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 - [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
|
20fc37778f |
fix(runner): keep settled OpenCode sessions across a controller proxy rebuild (#19641)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work.
> - The native Runner keeps an OpenCode conversation in durable state,
so a task can continue after a control-plane restart.
> - Saved OpenCode state binds one digest of the launch profile: the
Node command, the proxy bundle and the OpenCode executable, each by path
and SHA-256.
> - The Node command and the proxy bundle are controller runtime. A
rebuild or deploy between a hard kill and recovery replaces their bytes.
> - The recovery probe then refused a settled conversation with
`provider_state_incompatible`, although the OpenCode release and the
conversation were unchanged.
> - ACPX already re-binds a settled session after a Runner upgrade
(#18879). OpenCode did not.
> - This pull request records the OpenCode release digest in saved state
and re-binds a settled session when that release is unchanged.
> - The benefit is that a task continues automatically after a restart
that also changed the controller build. A real release change still
fails closed.
## Linked Issues or Issue Description
Refs #19558, #19501, #18879.
**What happened?**
A restart-chaos wave killed the control plane with SIGKILL while an
OpenCode run was settled between turns. The first recovery attempt
restored the session. The controller's runner `dist` was then rebuilt in
the same checkout. The next recovery attempt failed in the
`runner.drain` resume probe with `OpenCode runner launch profile changed
across durable recovery` (`native_provider_state_incompatible`). The
saved state matched the original proxy bytes exactly. Only the proxy
bundle bytes had changed. The OpenCode executable, the paths and the
model were the same.
**Expected behavior**
A settled OpenCode session continues after a restart when only
controller runtime bytes changed. A changed OpenCode release, or a
session with work in flight, fails closed. The existing disposition then
offers one Retry that continues in a fresh provider session.
**Steps to reproduce**
1. Start an OpenCode run and let it settle between turns.
2. Kill the runner and the controller.
3. Replace the qualified proxy bundle with valid new bytes (rebuild or
deploy).
4. Recover the run. The resume probe fails with
`provider_state_incompatible`.
**Paperclip version or commit**
`master` at
|
||
|
DottaandClaude Opus 5.5
|
e05bced39f |
fix(runner): adopt an accepted Codex turn whose start outlives the bound (#19608)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - Native Codex runs submit each turn through runnerd. The transport
answers `turn/start` only after it sees the provider's `turn/started`
for the turn the durable command result named
> - That wait has a fixed 30 s bound
> - At host load around 50, after two hard restarts, Codex accepted a
turn and started it about 27 s after the command, beyond the bound once
scheduling delay was added. The transport failed the start, but the
provider kept running the turn and made a tool call
> - No controller owned that turn. Its close could not settle, and the
task's Retry was refused at once
> - This pull request adopts the accepted turn when the bound expires,
but only when the runner still reports that turn active
> - The benefit is that a slow but live provider turn stays owned, so
its events, Stop and close keep working
## Linked Issues or Issue Description
**What happened?**
In a restart chaos replay (two SIGKILLs about 20 s apart, local runner,
real Codex, load 50), a resumed run failed with `runnerd did not report
the provider turn identity`. The runner journal shows the durable
`turn.start` result with the provider turn id, then `turn.started` for
that same turn 27 s later, followed by items and a semantic tool call.
The run's close then stayed incomplete (`semanticResultsSettled:
false`), and a Retry of the task failed with
`native_session_cleanup_quarantined: prior session cleanup remains
incomplete`.
**Steps to reproduce**
1. Use a Codex provider that answers `turn/start` and emits
`turn/started` later than the transport's turn-start bound.
2. Start a turn through the runnerd transport.
**Expected behavior**
The accepted turn is owned and observed by the controller.
**Actual behavior**
The start fails after the bound, while the provider still runs the turn.
## What Changed
- `runnerd-codex-transport.ts` (`#startTurnWork`):
- Near the end of the bound, if `turn/started` has not arrived, the
transport reads an authenticated `session.snapshot`. It reserves at most
5 s (or a quarter of the bound) for this read.
- It binds the accepted turn only when the turn is provably live:
- the runner reports that exact provider turn as active;
- the runner attributed no provider launch
(`provider_startup_ownership`) to a command after this `turn.start`;
- for a local provider, the same process (pid and start time) that
accepted the turn is still alive.
- Otherwise it keeps waiting for `turn/started` until the configured
bound and fails exactly as before. A turn lost across a provider
relaunch keeps the same-run resume path.
- `fake-codex-app-server.rs`: new test options `--turn-started-delay-ms`
(a delay between the `turn/start` response and `turn/started`) and
`--exit-before-turn-started`.
## Verification
- Red before the fix: `pnpm --filter @paperclipai/paperclip-runner exec
vitest run src/live/runnerd-codex-transport-deadlines.test.ts -t "adopts
a durably accepted"` (real runnerd and fake Codex, 12 s start delay, 8 s
bound) failed with `runnerd did not report the provider turn identity`.
This is the replay's error.
- Green after the fix: the start resolves with `provider-turn-1`, and
`turn/started` and the turn's terminal follow in order. There is one
provider `turn/start`.
- Negative case (`does not adopt an accepted turn the runner no longer
reports active`): the provider accepts the turn and exits before
starting it. The runner relaunches the provider and still lists the
turn, but the turn is not adopted. The start fails within the bound plus
scheduling slack. A snapshot-only version of this change adopted that
dead turn, and it hung for 20 s with no terminal.
- `src/native-session-runtime.test.ts`,
`src/live/runnerd-restart-continuation.test.ts` and
`src/live/runnerd-codex-transport-deadlines.test.ts` pass. The package
typecheck passes.
- Live chaos replays: results are added in a comment.
## Risks
- A provably live provider that accepts a turn and never starts it now
waits on the turn's own timeout and Stop paths instead of failing after
30 s. A dead or relaunched provider still fails at the bound.
- Remote runners can adopt only on the runner's own evidence (active
turn and no later provider launch), because their provider pid is not a
local process.
- A stale `turn/started` for a previous turn that arrives after the new
command result is still rejected. The transport cannot attribute that
turn's following events, so rejecting it fails only the transport, and
the same-run resume continues the task (seen in the replay). Delivering
it would attach the old turn's items and usage to the new turn.
## 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
Refs #19532, #19539.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
|
||
|
DottaandClaude Opus 5.5
|
785ead5c7a |
fix(tasks): hold a Retry while the task's own native session close is retained (#19609)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - When a native provider session close fails, the runner keeps a cleanup owner or quarantine in process memory. Its admission gate refuses new sessions for that task (scoped to the task since #19532) > - The server could not see that state. A task's Retry was accepted (202), its run was dispatched, and the runner refused it at once with `native_session_cleanup_quarantined: prior session cleanup remains incomplete` > - The user saw a Retry that failed instantly, and nothing said what to do > - This pull request lets the server read the runner's per-task cleanup gate and hold the Retry with a clear reason. The Retry starts automatically when the close is released > - The benefit is that a Retry is never dispatched only to fail on a block that is already known ## Linked Issues or Issue Description **What happened?** After two hard restarts, a Codex run's close stayed incomplete. The board Retry for that task was accepted, and its run failed in about 1 s with `native_session_cleanup_quarantined: prior session cleanup remains incomplete`. The next Retry did the same. **Steps to reproduce** 1. Let a native session close fail or stay pending for a task. 2. Press Retry on that task. **Expected behavior** The Retry waits with a clear reason, and starts when the cleanup finishes. If the cleanup needs an operator, Retry is not offered and the one working action is shown. **Actual behavior** The Retry is accepted and its run fails at once. ## What Changed - `native-session-runtime.ts`: - New `nativeSessionCleanupBlock({ companyId, issueId })`. It is a read-only view of the in-process admission gate for one task: `pending` (a retained close or a bounded automatic close recovery still owns the task's provider resources), `operator_required`, or `null`. It grants no authority. - New `onNativeSessionCleanupSettled(listener)`. It is called whenever a cleanup owner or quarantine entry is released. - `explicit-native-continuation.ts` (the continuation gate for a Retry or a new message): - `pending`: the input is held with `local_cleanup` ("Retry is waiting for the previous execution to finish cleanup. Your message is saved."). It starts automatically when the close is released. - `operator_required`: handled exactly like a recorded `native_session_cleanup_quarantined` run. Retry is not offered. A new message runs the existing stopped-process proof (`verifyStoppedNativeSessionForContinuation`), which retires the runner quarantine when the old processes are proven stopped. - `native-cleanup-release-wake.ts` (registered in `server/src/index.ts`): each cleanup release wakes deferred execution for the company in a transaction, so a held Retry re-evaluates. - `server/src/vendor/paperclip-runner/index.ts`: exports the two runtime functions. ## Verification - Red before the gate (server change stashed): `pnpm --filter @paperclipai/server exec vitest run src/services/quarantined-task-retry.test.ts -t "holds a Retry"` dispatched a successor run while the task's close was retained (`expected [ { … } ] to have a length of +0 but got 1`). - Green after the gate: - The Retry is held (`deferred_issue_execution`, `executionWait.reason = local_cleanup`), and no successor is dispatched. - A different task is not blocked. - When the close settles, the release is announced, and the next pass dispatches exactly one successor. - The whole `quarantined-task-retry.test.ts` file passes (28 tests). `src/native-session-runtime.test.ts` passes. Server `tsc --noEmit` passes after the runner package build. - Live chaos replays: results are added in a comment. ## Risks - The gate reads process memory, so it applies only where the server and the runner session share a process (local runner). Other deployments keep today's behavior. - A held Retry re-evaluates on the release wake (sent in its own transaction) or on the existing 30 s deferred-execution pass. When the bounded automatic close budget is spent with nothing else scheduled, the view reports `operator_required`. Retry is then no longer offered, and a new message reaches the stopped-process recovery instead of waiting forever. - No schema or API shape changes. ## 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 Refs #19532. 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com> |