From d6e6d1c78a5f3777a3ccfd350b5f769b6fa4c9f5 Mon Sep 17 00:00:00 2001
From: Dotta <34892728+cryppadotta@users.noreply.github.com>
Date: Sat, 10 Oct 2026 07:15:41 -0500
Subject: [PATCH] Stream OpenCode immediately, scope reasoning to turns, and
deduplicate instructions (#15708)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
## Thinking Path
> - Paperclip helps people manage AI agents and observe their work.
> - The native runner carries provider output into the task transcript.
> - OpenCode sends incremental text in `message.part.delta` events.
> - The driver ignored those events and waited for a full part snapshot.
Accepted chats also waited after lifecycle verification was complete.
> - This pull request forwards each valid chunk as it arrives and
removes unnecessary isolated startup work.
> - Valid chunks reach Paperclip sooner without changing completion or
permission rules. The matched browser experiment below does not
establish a reliable end-to-end latency improvement.
## Linked Issues or Issue Description
**What happened?**
The native OpenCode driver withheld text until OpenCode sent a full
snapshot. A local delayed-provider probe with the pinned OpenCode
1.18.34 runtime showed a 1,496 ms gap between the first native text
delta and the first Paperclip text delta (1,510 ms from provider text).
**Expected behavior**
Forward valid assistant text and reasoning deltas immediately. Preserve
repeated identical tokens, message roles, turn ownership, and final
response selection.
**Steps to reproduce**
Send an empty text part, then text deltas, and withhold the final full
snapshot. The new fixture requires the consumer to acknowledge its first
text delta before releasing the snapshot. The previous driver cannot
pass this handshake.
**Paperclip version or commit**
Baseline: `011e5669bad61c435a4745fb10812bef70376841`. The patch is
rebased onto current master.
**Deployment mode**
Local native OpenCode 1.18.34 on macOS arm64. Live diagnostics use
`openrouter/deepseek/deepseek-v4-flash-0731`. These are narrow
diagnostics, not full Product E2E qualification.
## What Changed
- Keep the macOS verified OpenCode launch pathname until child exit,
with guarded reuse and cleanup on failed launches. This fixes a
reproduced asynchronous executable-lifetime failure.
- Dispatch accepted queued chats immediately after lifecycle readiness
commits, through normal scheduler admission. Preserve ready state if
dispatch fails.
- Validate native OpenCode model access with the real hello probe
without a redundant catalog scan. Keep legacy adapter discovery
unchanged and isolate unspecified native probe directories.
- Map incremental text and reasoning to the existing canonical event
path, using observed part and message identities.
- Keep additive chunks without IDs distinct. Preserve ID-based event
deduplication.
- Clear current streaming part metadata when a new turn starts. Reject
mismatched session/message identities and unknown parts or fields.
- Disable automatic updates, remote model refreshes, and default plugins
in each isolated OpenCode process. Keep assigned MCP tools and default
model reasoning intact.
- Retain the latest cumulative text/reasoning part while its message
role is unknown, including completion metadata after more than 100
chunks.
- Expose `startTurn({ message, reasoningMode: "disabled" })` for one
OpenCode/OpenRouter turn. Carry the selection in the durable command
through Rust and the proxy. An omitted/default option restores provider
defaults on the next turn, including warm and recovered sessions. There
is no environment-level setting or automatic selection policy.
- Group optional Rust turn inputs in `CodexTurnOptions` and dispatch
through `start_turn_with_options`; retain `start_turn` for defaults.
Future turn settings can be added without another feature-specific
method.
- Expose a native-session capability for per-turn reasoning and reject
explicit unsupported selections before dispatch, including direct ACPX
and Codex backends. Runnerd checks the durable selected model as well as
the provider.
- Remove duplicate agent instructions from the shared OpenCode proxy
task envelope for all users. Send one system copy on every turn,
including prepared and resumed turns, because OpenCode reads the latest
user message’s system field. Preserve task constraints and completion
contracts.
- Document startup and streaming behavior. Add regression coverage for
role ordering, repeated chunks, final snapshot deduplication, and stale
turns.
## Verification
### Lower-load browser retry — 2026-10-10
**All 8/8 trials succeeded. In this batch, median Send → first visible
reply fell from 30.586 s to 12.476 s (59.2% lower).** Four fresh-agent
trials per variant, phase order before/after/after/before with two
trials per phase. Every adjacent-phase pair favored the candidate,
although one difference was only 0.301 s. This is evidence for the
readiness/probe changes on this host, not a production percentile or a
guarantee across providers.
| Phase/trial | Variant | Send → reply (s) | Observed ready → run (s) |
Session open → turn start (s) | Model requests |
|---|---|---:|---:|---:|---:|
| 0/0 | before | 45.042 | 25.337 | 3.629 | 2 |
| 0/1 | before | 13.689 | 0.000 | 2.443 | 2 |
| 1/0 | after | 14.696 | 0.000 | 2.526 | 1 |
| 1/1 | after | 13.389 | 0.000 | 2.470 | 2 |
| 2/0 | after | 10.871 | 0.000 | 2.706 | 2 |
| 2/1 | after | 11.564 | 0.004 | 2.443 | 2 |
| 3/0 | before | 41.582 | 21.518 | 2.376 | 2 |
| 3/1 | before | 19.589 | 9.753 | 2.707 | 1 |
Observed ready-to-run median: **15.636 s → 0 s**; zero means execution
began by the first ready observation, not literally zero latency.
Polling interval is nominally 250 ms plus HTTP time.
Session-open-to-turn-start includes controller handling, not just child
startup. All eight ran within the existing startup timeouts; no timeout
or reasoning policy was changed. Host load at fixture creation stayed
**5.47–8.73 on 18 CPUs**, versus 68.9–100.8 in the prior campaign. No
builds/tests ran concurrently with measurement.
Same merged source **b0a0158729a7913882a0e756fe7e27228d72bf53**, release
runner, rebuilt static UI, OpenCode **1.18.34**, model
**openrouter/deepseek/deepseek-v4-flash-0731**, provider-default
reasoning, and empty isolated workspaces in both variants. Only the
readiness worker and explicit-model verification source were toggled;
the launch-lifetime fix and all proxy streaming changes were held
constant. Thus this does not separately measure the original streaming
changes. One browser per phase, two fresh companies/agents/chats per
phase. Measurement begins at capture-phase Send click and ends on
visible Hi text plus two animation frames. Status/thinking is excluded;
emoji can arrive just after the first text. Successful screenshots
inspected. Each variant had three two-request turns using
`paperclip_finish` and one one-request turn; no status lookup was
observed. Reported usage includes settlement after the first visible
reply.
Runner SHA-256
`57f658c220eecd2370ffd3333cd7e5d06d16bb2798af0cf43c84e01c8207c1c3`;
proxy
`bcc02fda2bf35b38a116cd880841aa6c534f6e0b497a96d97369a1e06dcc72ac`;
rebuilt UI tree
`eee238f2eda628c4e8c3660707986da4f8dd9100c814e26938fce3756df2eae4`
(sorted relative path, NUL, binary per-file SHA-256). Worker/probe
variant source hashes are unchanged from the preserved comparison below.
Measured turn cost **$0.00612574**, plus verification probes not
included in that figure. All eight attempts retained; all source/proxy
files restored and test supervisors exited.
Current commit validation: **53 successful CI checks, 4 skipped, zero
failing; Greptile 5/5 on b0a0158729 with no actionable findings.** The
previous 20 targeted post-merge tests passed. No source edit was made
for this rerun. Earlier failed and inconclusive cohorts below remain
historical evidence and are not pooled with this batch.
### Master update and failure diagnosis — 2026-10-09
Merged `origin/master` at `417db14e7d` into this branch without
conflicts; pushed merge commit
`b0a0158729a7913882a0e756fe7e27228d72bf53`. The incoming changes do not
alter OpenCode process startup or its timeout budgets. Historical
measurements below remain attributed to their original commits.
The four last-campaign failures never issued `turn.start`, so these were
**cold-session admission failures, not model replies failing**. Two
retained `session.open` commands eventually completed after controller
timeout/cleanup had begun. The other two ended with the inner
`thread/start` response timeout, and their diagnostics recorded OpenCode
startup retries. Startup event timestamps are buffered publication times
and must not be interpreted as exact process-spawn times.
On the merged source, two credential-free startup-only checks used the
real pinned OpenCode 1.18.34, empty temporary workspaces, a placeholder
key, and **no model turns**:
- Existing pinned executable: spawn at 1.083 s; failed health
attempt/retry at 11.311 s and 21.914 s; third process healthy at 32.210
s; session opened at **35.168 s**.
- Verified private executable path: snapshot preparation took **0.497
s**; process spawned at 1.588 s; healthy at 9.967 s; session opened at
**11.120 s**.
These are sequential diagnostic samples, not evidence that one
executable path is faster. Host load was **56.5–64.5 on 18 logical
CPUs**. The local health wait permits 10 s per attempt and the driver
retries up to three times; the outer controller's cold admission budget
is 30 s, and the Rust facade also waits 30 s for a response. Thus
retries plus setup/session creation can outlast the caller's budget. The
isolated 35.168 s success demonstrates that a session can eventually
open yet fail the app's admission deadline. Heavy load aggravates the
timing; its exact contribution is not isolated. No timeout was increased
and no new model-latency claim is made.
Merge validation: **20/20 targeted lifecycle-policy and heartbeat
run-completion tests passed**, including temporary PostgreSQL cases. The
first sandbox invocation skipped database cases; the authorized local-DB
rerun executed all 20. The package-manager launcher could not verify its
version via the unreachable registry, so validation used the
already-installed Vitest binary. Full CI/review subsequently passed on
this merge commit; see the October 10 retry above. Older results below
apply to their named commits.
### Follow-up: reproduced macOS launch failure — 2026-10-09
The retry exposed a startup bug rather than only host-load noise. The
macOS proxy deleted the private verified executable path immediately
after asynchronous spawn. A credential-free test of the actual pinned
OpenCode `--version` reproduced a fast SIGKILL and a 15-second timeout
in three immediate-cleanup attempts; three deferred-cleanup controls
succeeded. The timeout was killed by the diagnostic's own deadline. This
experiment does not establish the OS-internal reason for the fast
SIGKILL.
Commit `c30451e0ebf3f74e9ef70d37bbe4d9fe7fb56359` retains the private
launch path until child exit, cleans failed spawns, and waits for forced
shutdown before reuse. Source descriptor authentication, ownership,
mode, inode checks, and the prohibition on ambient executable fallback
remain. Overlapping reuse fails closed. The existing synchronous shell
test was changed to exercise the actual asynchronous boundary.
**Verification:** 59 focused proxy/driver tests passed; runner
TypeScript check passed. All six subsequent real OpenCode version
launches succeeded without timeout (4.11, 4.33, 5.73, 6.91, 4.95, 6.70
seconds). The native Rust binary is unchanged from the retry; no model
calls were made by these diagnostics.
Retained interrupted retry: before 42.56/18.69 seconds; after one
180-second timeout (`provider_process_exited` / health SIGKILL) and one
37.33-second success. The latter overlapped the credential-free
diagnostic and is not a clean matched timing. The next phase was stopped
before Send to diagnose the reproduced bug. Reported measured-turn cost:
$0.0016874, verification probes additional. These attempts are not
pooled with other campaigns.
**Completed retry, stopped by the declared failure rule after 6 of 8
planned trials. No reliable end-to-end speedup demonstrated.** At commit
`c30451e0ebf3f74e9ef70d37bbe4d9fe7fb56359`, the launch fix above was
held constant in both variants; only the readiness handoff and
explicit-model verification path changed. Planned phase order:
before/after/after/before, two fresh agents/chats per phase, one new
browser per phase. The second after phase had two consecutive failures,
so the final before phase was not run.
| Phase/trial | Variant | Send → first visible reply | Observed ready →
run | Outcome |
|---|---|---|---|---|
| 0/0 | before | No reply within 180 s | 2.777 s | session.open timeout
|
| 0/1 | before | 47.654 s | 9.383 s | Succeeded |
| 1/0 | after | 50.051 s | 0.087 s | Succeeded |
| 1/1 | after | No reply within 180 s | 0.151 s | session.open timeout |
| 2/0 | after | No reply within 180 s | 3.337 s | session.open timeout |
| 2/1 | after | No reply within 180 s | 1.434 s | session.open timeout |
Both successful screenshots show “Hi 👋”. Timing starts at the browser's
capture-phase Send click and ends at the first visible answer text plus
two animation frames; the first fragment can be “Hi” before the emoji
arrives. Thinking/status text is not counted. All four failures were
`runner_local_connect_failed: PRP command session.open timed out`
(`native_session_interrupted`), before a reply; none reported the
earlier health-stage SIGKILL. They are retained, not discarded as
outliers. One success per variant does not support a speedup estimate or
median comparison. Observed ready-to-run median was **6.080 s before
(n=2) vs 0.793 s after (n=4)**; this is sampled every nominal 250 ms
plus HTTP latency and is a lower bound, not precise tracing. Host load
at fixture creation ranged **68.9–100.8 on 18 logical CPUs**, a
substantial confound that worsened over the campaign. No local
build/test was run concurrently with this campaign.
Provenance: pinned OpenCode **1.18.34**, model
**openrouter/deepseek/deepseek-v4-flash-0731**, default reasoning in
both variants, empty synthetic workspaces and isolated probe homes.
Release runner SHA-256
`57f658c220eecd2370ffd3333cd7e5d06d16bb2798af0cf43c84e01c8207c1c3`;
current proxy SHA-256
`bcc02fda2bf35b38a116cd880841aa6c534f6e0b497a96d97369a1e06dcc72ac`;
built UI tree SHA-256
`7f0c2b7f586cb753408351574b2239e22311944add99af4f6f3562cd6d47bbcb`. This
holds the rebuilt runner and static UI constant. Provider-reported
measured turn cost **$0.00120649**; verification probes are additional.
Including the earlier interrupted retry, this follow-up reports
**$0.00289389** in measured turn usage; missing usage is not assumed
free. Prior cohorts remain separate below.
Validation on `c30451e0e`: **59 focused tests passed**, runner
TypeScript check passed, **6/6 real credential-free pinned OpenCode
launches passed**, fresh Greptile **5/5 with zero unresolved threads**.
CI: **51 success, 4 skipped, 1 failed**. The failed native compilation
job never reached compilation: Docker's auth endpoint timed out during
Buildx setup, then failed the same way on one retry
([job](https://github.com/paperclipai/paperclip/actions/runs/37994181674/job/114039737503)).
Full local checks were not rerun while measuring. The PR remains draft:
the launch defect is reproduced and fixed, but cold session startup
still exceeds the controller's admission budget on this host. Raising
that budget would not establish a faster first response.
### Readiness investigation and live rerun — 2026-10-09
**Measured queue fix; no reliable end-to-end speedup claim. Keep this PR
in draft.** Accepted chats could wait for the periodic queue sweep after
verification had already committed `ready`. The lifecycle worker now
dispatches through the existing scheduler immediately after that commit.
The normal company, pause, budget, concurrency, and task-ownership
checks still apply. Dispatch failure preserves readiness and leaves
recovery to the queue sweep.
Native OpenCode verification also avoids enumerating the full model
catalog before running its real hello request. Identifier validation
remains. Authentication failures, unavailable models, and timeouts still
prevent readiness. An unspecified local probe directory uses an empty
temporary directory.
**Queue-only experiment:** four before and four after attempts, in
before/after/after/before server phases, two fresh agents per phase.
Median observed ready-to-start delay fell from **19.429 s to 0.058 s**.
State polling was nominally 250 ms plus HTTP time, so the latter means
within observation resolution; it is not a precise 58 ms latency claim.
After timings were 100.20, 65.29, 63.78, and 76.97 seconds. Before had
one 59.65-second success and three 120-second response timeouts (one
provider health SIGKILL, one late success, one still running after
settlement and canceled). These censored results do not support a
success-only median speedup comparison. Focused tests overlapped the
last baseline phase, which further limits total-time interpretation.
Reported measured-turn cost: $0.0055383; verification probes and
incomplete/canceled usage are additional.
**Final matched experiment:** built static UI in both variants to avoid
pre-Send Vite compilation; current proxy, release runner, selected
model, default reasoning, and explicitly isolated empty probe
cwd/HOME/XDG held constant. Both worker handoff and model-probe changes
vary. Fresh browser/company/agent/chat per attempt. ABBA order, two
attempts per variant, 180-second response limit declared before the
batch, four-turn bound. Same Send-to-rendered-Hi metric as the earlier
experiment.
| Order | Variant | Send → visible Hi | Created → execution | Observed
ready → execution | Outcome |
| --- | --- | ---: | ---: | ---: | --- |
| 1 | Before | 62.66 s | 23.467 s | 2.830 s | Succeeded |
| 2 | After | >180 s | 14.200 s | Within polling resolution | Provider
health SIGKILL |
| 3 | After | >180 s | 13.877 s | Within polling resolution | Provider
initialization timeout |
| 4 | Before | >180 s | 144.256 s | 14.067 s | Still running after
settlement; canceled |
Only one of four attempts produced a timed reply. **There is no valid
overall speedup estimate from this batch.** The dispatch delay is
removed, but provider startup failures and heavy shared-host load
remain. A recorded host sample was load 93.55 on 18 logical CPUs. This
is a confound, not proof that every failure was caused by host load.
Provider-reported measured-turn cost was $0.00032752, with no reported
usage for the other three attempts; verification probes are additional.
Retained earlier setup attempts: one shared-memory startup failure
before model calls, and a separate development-UI batch with a
60.06-second baseline followed by a navigation timeout before Send.
These attempts are retained separately, not counted as successful
response timings. No live attempts were silently dropped.
Provenance: source base `53580f8fbd40b963f3fb7ba5172ff3ede2e9ef1f`,
measured patch committed as `40b54eafb`, then rebased as
`c590770dcc2bb1debce14037935a7031e5c3c1ac`. **These live timings predate
the rebase.** Source and binary hashes are retained below. The inner
harness label `after` selects the identical current proxy in every
attempt; the outer campaign handoff label is the actual before/after
variant. Owned benchmark processes stopped and original proxy restored.
Post-rebase validation at `c590770dcc2bb1debce14037935a7031e5c3c1ac`:
- Full local `pnpm -r typecheck`: **passed**, including the native
release build (1,098.6 seconds on the loaded host).
- Focused lifecycle and adapter tests: **26 passed**. The 14 database
integration tests did not run locally because their hard-coded 20-second
database setup hook timed out. A retry with a CLI timeout override
encountered the same explicit hook limit. The earlier pre-rebase focused
batch passed 77 tests; these results are not substituted for post-rebase
execution.
- Local `pnpm test:run` was started, then stopped to avoid duplicating
CI on the saturated host. Local `pnpm build` was not run separately.
**Current-head CI passed Typecheck + Release Registry, Build, all
general and serialized test groups, Rust checks, both runner Vitest
groups, and all eight browser E2E shards.** The last general shard
passed 1,179 tests (72 files).
- Fresh [Greptile
review](https://github.com/paperclipai/paperclip/pull/15708#issuecomment-6085742217):
**5/5**, reviewed this exact head, no actionable defect.
- Final CI rollup: **49 success, 4 skipped, 4 failure**. The failures
are [Compile isolated native
Runner](https://github.com/paperclipai/paperclip/actions/runs/37990274515/job/114024329242),
[Docker context
integrity](https://github.com/paperclipai/paperclip/actions/runs/37990274623/job/114024196511),
[Canary Dry
Run](https://github.com/paperclipai/paperclip/actions/runs/37990274623/job/114024196424),
and the dependent `ci / verify` aggregate. All three root failures are
Docker Hub HTTP 429 / unauthenticated pull-rate limits while fetching
Node base images. The native-image job was retried once and hit the same
limit. No source compiler/test failure was reported by those gates. CI
maintainers need authenticated pulls or to rerun after the registry
limit clears.
- Owned benchmark servers stopped. Local validation commands completed
or were stopped. The PR remains a draft: user-visible speed/reliability
is not qualified, and the Docker-dependent CI gates are not green.
Sanitized readiness evidence with all attempts and
source hashes
```json
{"campaign":{"order":["before","after","after","before"],"trialsPerPhase":1,"maxMeasuredTurns":4,"maxCampaignUsd":1,"sourceHead":"53580f8fbd40b963f3fb7ba5172ff3ede2e9ef1f","reasoning":"provider default, no override","proxySha256":"6691475ed55f10179ec6ee0023a1f6462982f763848da7cd5b1d0baf38633b04","purpose":"Final matched comparison of both optimizations; distinct from earlier queue-only campaign.","responseTimeoutMs":180000,"timeoutRationale":"Declared before this final campaign because earlier unchanged baselines sometimes exceeded 120 seconds; earlier timeouts remain retained.","ui":{"mode":"built static UI","treeSha256":"7f0c2b7f586cb753408351574b2239e22311944add99af4f6f3562cd6d47bbcb"},"priorCampaign":"combined-proof stopped after a development-UI navigation timeout before Send; no failed attempts erased."},"complete":true,"statistics":{"before":{"attempted":2,"successfulMeasurements":1,"medianTtfrMs":62657.2000002861,"rangeTtfrMs":[62657.2000002861,62657.2000002861],"medianObservedReadyToRunMs":8448.5,"medianRunStartToVisibleMs":37290.199951171875},"after":{"attempted":2,"successfulMeasurements":0,"medianTtfrMs":null,"rangeTtfrMs":null,"medianObservedReadyToRunMs":0.0,"medianRunStartToVisibleMs":null}},"measuredTurnCostUsd":0.00032752,"costCoverage":"Provider-reported measured chat turns only; harness verification probes are additional and not included.","readinessMetricLimit":"First observed ready state sampled every 250 ms plus HTTP duration; clamped to zero if execution started before the observation. This is a lower bound on ready-to-start delay, not a precise sub-millisecond metric.","results":[{"phase":0,"trial":0,"handoff":"before","ttfrMs":62657.2000002861,"answer":"Hi","statuses":["succeeded"],"error":null,"runErrorCode":null,"providerHealthKilled":false,"queueMs":23467.0,"queuedBeforeObservedReady":true,"observedVerificationMsAfterFixture":26390.0,"observedReadyToRunMs":2830.0,"readyAndStartedInSameSample":false,"runStartToVisibleMs":37290.199951171875,"usage":[{"model":"openrouter/deepseek/deepseek-v4-flash-0731","costUsd":0.00032752,"costUsdExact":"0.000327520","inputTokens":26864,"outputTokens":46,"cachedInputTokens":0,"cacheWriteTokens":0,"usageSource":"per_run","costStatus":"reported"}],"workerSha256":"02adc7e1162fb60439869cec62b75d8fb22fec71155f80535c8e00211cde8326"},{"phase":1,"trial":0,"handoff":"after","ttfrMs":null,"answer":null,"statuses":["failed"],"error":"TimeoutError: page.waitForFunction: Timeout 180000ms exceeded.","runErrorCode":"provider_process_exited","providerHealthKilled":true,"queueMs":14200.0,"queuedBeforeObservedReady":true,"observedVerificationMsAfterFixture":24028.0,"observedReadyToRunMs":0,"readyAndStartedInSameSample":true,"runStartToVisibleMs":null,"usage":[{}],"workerSha256":"ceffb90a6bc97ffaaf8ec7d0267c7bfa7738d321aaf2a618b5efb9b114bc88e7"},{"phase":2,"trial":0,"handoff":"after","ttfrMs":null,"answer":null,"statuses":["failed"],"error":"TimeoutError: page.waitForFunction: Timeout 180000ms exceeded.","runErrorCode":"provider_initialize_timeout","providerHealthKilled":false,"queueMs":13877.0,"queuedBeforeObservedReady":true,"observedVerificationMsAfterFixture":18107.0,"observedReadyToRunMs":0,"readyAndStartedInSameSample":true,"runStartToVisibleMs":null,"usage":[{}],"workerSha256":"ceffb90a6bc97ffaaf8ec7d0267c7bfa7738d321aaf2a618b5efb9b114bc88e7"},{"phase":3,"trial":0,"handoff":"before","ttfrMs":null,"answer":null,"statuses":["running"],"error":"TimeoutError: page.waitForFunction: Timeout 180000ms exceeded.","runErrorCode":null,"providerHealthKilled":false,"queueMs":144256.0,"queuedBeforeObservedReady":true,"observedVerificationMsAfterFixture":136901.0,"observedReadyToRunMs":14067.0,"readyAndStartedInSameSample":false,"runStartToVisibleMs":null,"usage":[{}],"workerSha256":"02adc7e1162fb60439869cec62b75d8fb22fec71155f80535c8e00211cde8326"}],"hostLoadSample":{"at":"2026-10-09T20:48:40.031Z","parallelism":18,"load":[93.546875,82.68017578125,75.9091796875]},"measuredChangesCommit":"40b54eafb","postRebaseCommit":"c590770dcc2bb1debce14037935a7031e5c3c1ac","postRebaseLimit":"Live benchmark predates the rebase. No post-rebase live result is claimed.","variantSourceHashes":{"before":{"server/src/modules/agent-lifecycle/application/worker.ts":"02adc7e1162fb60439869cec62b75d8fb22fec71155f80535c8e00211cde8326","packages/adapters/opencode-local/src/server/test.ts":"4c10bacb55f2c7589e7b5c79d74781206ae075db8955f5b998065d6db3c0335a","server/src/adapters/registry.ts":"4a858763f69947d5745c7b8cb251a62fa404bc9683ade0ffb76282ed12761394"},"after":{"server/src/modules/agent-lifecycle/application/worker.ts":"ceffb90a6bc97ffaaf8ec7d0267c7bfa7738d321aaf2a618b5efb9b114bc88e7","packages/adapters/opencode-local/src/server/test.ts":"cd2ea3d51c3f1b4fb42261be701fb7f6b85704ad8e641e6db73d7c58e47364d3","server/src/adapters/registry.ts":"4a858763f69947d5745c7b8cb251a62fa404bc9683ade0ffb76282ed12761394"}},"model":"openrouter/deepseek/deepseek-v4-flash-0731","prompt":"Reply with Hi and an emoji and nothing else.","runnerBinary":{"profile":"release","sha256":"6134c17debda14a76da69ac9631a2d10dad482ebd58dbc9a4e4480ed652211c4"},"probeIsolation":"Explicit empty cwd and HOME/XDG folders under the synthetic fixture workspace in BOTH variants. No server checkout or ambient provider home is used for the hello request.","comparisonLabelNote":"Inner run.mts variant after selects the identical current proxy in every trial; outer campaign readinessHandoff controls before/after source changes."}
```
Queue-only campaign:
```json
{"campaign":{"order":["before","after","after","before"],"trialsPerPhase":2,"maxMeasuredTurns":8,"maxCampaignUsd":1,"sourceHead":"53580f8fbd40b963f3fb7ba5172ff3ede2e9ef1f","reasoning":"provider default, no override","proxySha256":"6691475ed55f10179ec6ee0023a1f6462982f763848da7cd5b1d0baf38633b04"},"complete":true,"statistics":{"before":{"attempted":4,"successfulMeasurements":1,"medianTtfrMs":59654.5,"rangeTtfrMs":[59654.5,59654.5],"medianObservedReadyToRunMs":19429.0,"medianRunStartToVisibleMs":8843.5},"after":{"attempted":4,"successfulMeasurements":4,"medianTtfrMs":71128.7000002861,"rangeTtfrMs":[63778.90000009537,100195.59999990463],"medianObservedReadyToRunMs":58.0,"medianRunStartToVisibleMs":26542.849975585938}},"measuredTurnCostUsd":0.0055383,"costCoverage":"Provider-reported measured chat turns only; harness verification probes are additional and not included.","readinessMetricLimit":"First observed ready state sampled every 250 ms plus HTTP duration; clamped to zero if execution started before the observation. This is a lower bound on ready-to-start delay, not a precise sub-millisecond metric.","results":[{"phase":0,"trial":0,"handoff":"before","ttfrMs":59654.5,"answer":"Hi 👋","statuses":["succeeded"],"error":null,"queueMs":50590.0,"observedReadyToRunMs":25709.0,"readyAndStartedInSameSample":false,"runStartToVisibleMs":8843.5,"usage":[{"model":"openrouter/deepseek/deepseek-v4-flash-0731","costUsd":0.00087852,"costUsdExact":"0.000878520","inputTokens":25644,"outputTokens":264,"cachedInputTokens":28416,"cacheWriteTokens":0,"usageSource":"per_run","costStatus":"reported"}],"workerSha256":"02adc7e1162fb60439869cec62b75d8fb22fec71155f80535c8e00211cde8326"},{"phase":0,"trial":1,"handoff":"before","ttfrMs":null,"answer":null,"statuses":["failed"],"error":"TimeoutError: page.waitForFunction: Timeout 120000ms exceeded.","queueMs":41651.0,"observedReadyToRunMs":14656.0,"readyAndStartedInSameSample":false,"runStartToVisibleMs":null,"usage":[{}],"workerSha256":"02adc7e1162fb60439869cec62b75d8fb22fec71155f80535c8e00211cde8326"},{"phase":1,"trial":0,"handoff":"after","ttfrMs":100195.59999990463,"answer":"Hi 🙂","statuses":["succeeded"],"error":null,"queueMs":65893.0,"observedReadyToRunMs":116.0,"readyAndStartedInSameSample":false,"runStartToVisibleMs":33243.60009765625,"usage":[{"model":"openrouter/deepseek/deepseek-v4-flash-0731","costUsd":0.00075782,"costUsdExact":"0.000757820","inputTokens":27014,"outputTokens":175,"cachedInputTokens":26368,"cacheWriteTokens":0,"usageSource":"per_run","costStatus":"reported"}],"workerSha256":"ceffb90a6bc97ffaaf8ec7d0267c7bfa7738d321aaf2a618b5efb9b114bc88e7"},{"phase":1,"trial":1,"handoff":"after","ttfrMs":65288.60000038147,"answer":"Hi 👋","statuses":["succeeded"],"error":null,"queueMs":43511.0,"observedReadyToRunMs":0,"readyAndStartedInSameSample":true,"runStartToVisibleMs":21227.60009765625,"usage":[{"model":"openrouter/deepseek/deepseek-v4-flash-0731","costUsd":0.00085326,"costUsdExact":"0.000853260","inputTokens":25166,"outputTokens":250,"cachedInputTokens":28160,"cacheWriteTokens":0,"usageSource":"per_run","costStatus":"reported"}],"workerSha256":"ceffb90a6bc97ffaaf8ec7d0267c7bfa7738d321aaf2a618b5efb9b114bc88e7"},{"phase":2,"trial":0,"handoff":"after","ttfrMs":63778.90000009537,"answer":"Hi 👋","statuses":["succeeded"],"error":null,"queueMs":39218.0,"observedReadyToRunMs":0,"readyAndStartedInSameSample":true,"runStartToVisibleMs":23479.89990234375,"usage":[{"model":"openrouter/deepseek/deepseek-v4-flash-0731","costUsd":0.000839,"costUsdExact":"0.000839000","inputTokens":27068,"outputTokens":238,"cachedInputTokens":26368,"cacheWriteTokens":0,"usageSource":"per_run","costStatus":"reported"}],"workerSha256":"ceffb90a6bc97ffaaf8ec7d0267c7bfa7738d321aaf2a618b5efb9b114bc88e7"},{"phase":2,"trial":1,"handoff":"after","ttfrMs":76968.80000019073,"answer":"Hi 👋","statuses":["succeeded"],"error":null,"queueMs":46257.0,"observedReadyToRunMs":185.0,"readyAndStartedInSameSample":false,"runStartToVisibleMs":29605.800048828125,"usage":[{"model":"openrouter/deepseek/deepseek-v4-flash-0731","costUsd":0.00030582,"costUsdExact":"0.000305820","inputTokens":24694,"outputTokens":32,"cachedInputTokens":1792,"cacheWriteTokens":0,"usageSource":"per_run","costStatus":"reported"}],"workerSha256":"ceffb90a6bc97ffaaf8ec7d0267c7bfa7738d321aaf2a618b5efb9b114bc88e7"},{"phase":3,"trial":0,"handoff":"before","ttfrMs":null,"answer":null,"statuses":["succeeded"],"error":"TimeoutError: page.waitForFunction: Timeout 120000ms exceeded.","queueMs":100979.0,"observedReadyToRunMs":3872.0,"readyAndStartedInSameSample":false,"runStartToVisibleMs":null,"usage":[{"model":"openrouter/deepseek/deepseek-v4-flash-0731","costUsd":0.00190388,"costUsdExact":"0.001903880","inputTokens":25524,"outputTokens":852,"cachedInputTokens":55808,"cacheWriteTokens":0,"usageSource":"per_run","costStatus":"reported"}],"workerSha256":"02adc7e1162fb60439869cec62b75d8fb22fec71155f80535c8e00211cde8326"},{"phase":3,"trial":1,"handoff":"before","ttfrMs":null,"answer":null,"statuses":["running"],"error":"TimeoutError: page.waitForFunction: Timeout 120000ms exceeded.","queueMs":129391.0,"observedReadyToRunMs":24202.0,"readyAndStartedInSameSample":false,"runStartToVisibleMs":null,"usage":[{}],"workerSha256":"02adc7e1162fb60439869cec62b75d8fb22fec71155f80535c8e00211cde8326"}]}
```
Sanitized measurement records, source/binary hashes, usage, and
retained attempt classification
```json
{"manifest":{"candidate":"53580f8fbd40b963f3fb7ba5172ff3ede2e9ef1f","baseline":"41fec8c8921f9fb19c13fe11e2c9b71bf1017317","comparison":"Same current product and runner; replace the three OpenCode proxy/driver source modules with exact pre-PR contents for baseline. Reasoning uses defaults in both.","model":"openrouter/deepseek/deepseek-v4-flash-0731","prompt":"Reply with Hi and an emoji and nothing else.","samplesPerVariant":6,"order":["before","after","after","before","before","after","after","before","before","after","after","before"],"maxTurnMs":120000,"maxCostUsd":1,"variants":{"before":{"sources":{"src/cli/opencode-app-server-proxy.ts":"e9b46c937d7ce3b51dc44181a2a3ad82befda9ebe184adf80c0a6b30da140fd6","src/cli/opencode-proxy-task-envelope.ts":"fc23f8a515c2c4d1551f6b95e5b38e6254849dec23f7396e76b646e170a6e006","src/drivers/opencode/opencode-server-driver.ts":"99de415e8fc3bb5fb6f8d7512d751c4c37d2aa77a18b0546b9fb232f1a6fb6c5"},"bundleSha256":"6dd8726e72545341f6772e54ec5aeb6264dec74db67290df44930d87135efb6b"},"after":{"sources":{"src/cli/opencode-app-server-proxy.ts":"fa800e9bccce02f05f1af83e225d3b5a1300d0a7d4c35ac548f400d6c0a83f79","src/cli/opencode-proxy-task-envelope.ts":"5c387614fc9832322686326b8482eb00d88699d491b05cf59cadcd94a41194ea","src/drivers/opencode/opencode-server-driver.ts":"a944f3bf7f1bf897daec3a5dc2aa34ee9d1a2133efcb3cf150b002095a0ec586"},"bundleSha256":"6691475ed55f10179ec6ee0023a1f6462982f763848da7cd5b1d0baf38633b04"}},"runnerBinary":{"profile":"release","sha256":"6134c17debda14a76da69ac9631a2d10dad482ebd58dbc9a4e4480ed652211c4"},"metric":"Browser Send capture-phase click to visible Hi answer, sampled on animation frames, plus two paint opportunities. Excludes reasoning/tool/status text.","setupAttempts":["setup-failure-1","setup-failure-2","startup-failure-3","pilot-selector-failure"]},"summary":{"before":{"attempts":6,"measured":6,"medianMs":48083.050000190735,"minMs":37420.5,"maxMs":66397.69999980927,"totalCostUsd":0.004945229,"toolCalls":[4,1,0,0,0,1]},"after":{"attempts":6,"measured":6,"medianMs":42504.049999952316,"minMs":27680.099999904633,"maxMs":63320.90000009537,"totalCostUsd":0.00557214,"toolCalls":[0,1,0,4,0,1]},"medianReductionPercent":11.602841334350234,"medianPairedImprovementMs":-168.59999990463257,"reportedCampaignCostUsdIncludingPilot":0.011371498999999998},"pairs":[{"indices":[0,1],"beforeMs":66397.69999980927,"afterMs":27680.099999904633,"improvementMs":38717.59999990463},{"indices":[2,3],"beforeMs":37420.5,"afterMs":27997,"improvementMs":9423.5},{"indices":[4,5],"beforeMs":39751.60000038147,"afterMs":55217.199999809265,"improvementMs":-15465.599999427795},{"indices":[6,7],"beforeMs":47318.90000009537,"afterMs":63320.90000009537,"improvementMs":-16002.0},{"indices":[8,9],"beforeMs":48847.2000002861,"afterMs":29790.900000095367,"improvementMs":19056.300000190735},{"indices":[10,11],"beforeMs":51311.7000002861,"afterMs":61072.40000009537,"improvementMs":-9760.699999809265}],"trials":[{"index":0,"variant":"before","startedAt":"2026-10-09T19:15:23.172Z","measurementError":null,"ttfrMs":66397.69999980927,"domMs":66390.59999990463,"answer":"Hi 👋","runCount":1,"statuses":["succeeded"],"usage":[{"model":"openrouter/deepseek/deepseek-v4-flash-0731","costUsd":0.002612595,"costUsdExact":"0.002612595","inputTokens":55962,"outputTokens":1344,"cachedInputTokens":83456,"cacheWriteTokens":0,"usageSource":"per_run","costStatus":"reported"}],"bundleSha256":"6dd8726e72545341f6772e54ec5aeb6264dec74db67290df44930d87135efb6b","toolCalls":["paperclip_get_task_context","paperclip_comment_on_task","paperclip_finish","paperclip_finish"],"eventCounts":{"turn.started":1,"tool.execution.started":4,"item.delta":1,"turn.completed":1},"eventTimes":{"session.started":"2026-10-09T19:15:58.194Z","turn.submitted":"2026-10-09T19:15:58.311Z","turn.started":"2026-10-09T19:15:59.063Z","firstTextDelta":"2026-10-09T19:16:36.039Z","turn.completed":"2026-10-09T19:16:36.118Z"},"completedRunDurationsMs":[42915.0]},{"index":1,"variant":"after","startedAt":"2026-10-09T19:16:36.598Z","measurementError":null,"ttfrMs":27680.099999904633,"domMs":27636.599999904633,"answer":"Hi","runCount":1,"statuses":["succeeded"],"usage":[{"model":"openrouter/deepseek/deepseek-v4-flash-0731","costUsd":0.00030429,"costUsdExact":"0.000304290","inputTokens":26845,"outputTokens":28,"cachedInputTokens":0,"cacheWriteTokens":0,"usageSource":"per_run","costStatus":"reported"}],"bundleSha256":"6691475ed55f10179ec6ee0023a1f6462982f763848da7cd5b1d0baf38633b04","toolCalls":[],"eventCounts":{"turn.started":1,"item.delta":3,"turn.completed":1},"eventTimes":{"session.started":"2026-10-09T19:17:00.717Z","turn.submitted":"2026-10-09T19:17:00.841Z","turn.started":"2026-10-09T19:17:01.508Z","firstTextDelta":"2026-10-09T19:17:05.934Z","turn.completed":"2026-10-09T19:17:06.205Z"},"completedRunDurationsMs":[14435.0]},{"index":2,"variant":"after","startedAt":"2026-10-09T19:17:08.762Z","measurementError":null,"ttfrMs":27997,"domMs":27984.599999904633,"answer":"Hi 👋","runCount":1,"statuses":["succeeded"],"usage":[{"model":"openrouter/deepseek/deepseek-v4-flash-0731","costUsd":0.00101016,"costUsdExact":"0.001010160","inputTokens":27544,"outputTokens":366,"cachedInputTokens":26624,"cacheWriteTokens":0,"usageSource":"per_run","costStatus":"reported"}],"bundleSha256":"6691475ed55f10179ec6ee0023a1f6462982f763848da7cd5b1d0baf38633b04","toolCalls":["paperclip_finish"],"eventCounts":{"turn.started":1,"tool.execution.started":1,"item.delta":1,"turn.completed":1},"eventTimes":{"session.started":"2026-10-09T19:17:26.192Z","turn.submitted":"2026-10-09T19:17:26.305Z","turn.started":"2026-10-09T19:17:26.936Z","firstTextDelta":"2026-10-09T19:17:40.688Z","turn.completed":"2026-10-09T19:17:40.826Z"},"completedRunDurationsMs":[17763.0]},{"index":3,"variant":"before","startedAt":"2026-10-09T19:17:41.376Z","measurementError":null,"ttfrMs":37420.5,"domMs":37364.40000009537,"answer":"Hi 👋","runCount":1,"statuses":["succeeded"],"usage":[{"model":"openrouter/deepseek/deepseek-v4-flash-0731","costUsd":0.000693018,"costUsdExact":"0.000693018","inputTokens":24236,"outputTokens":282,"cachedInputTokens":27648,"cacheWriteTokens":0,"usageSource":"per_run","costStatus":"reported"}],"bundleSha256":"6dd8726e72545341f6772e54ec5aeb6264dec74db67290df44930d87135efb6b","toolCalls":["paperclip_finish"],"eventCounts":{"turn.started":1,"tool.execution.started":1,"item.delta":1,"turn.completed":1},"eventTimes":{"session.started":"2026-10-09T19:18:12.381Z","turn.submitted":"2026-10-09T19:18:12.616Z","turn.started":"2026-10-09T19:18:13.441Z","firstTextDelta":"2026-10-09T19:18:21.746Z","turn.completed":"2026-10-09T19:18:22.262Z"},"completedRunDurationsMs":[29216.0]},{"index":4,"variant":"before","startedAt":"2026-10-09T19:18:23.000Z","measurementError":null,"ttfrMs":39751.60000038147,"domMs":39705.7000002861,"answer":"Hi 👋","runCount":1,"statuses":["succeeded"],"usage":[{"model":"openrouter/deepseek/deepseek-v4-flash-0731","costUsd":0.000265747,"costUsdExact":"0.000265747","inputTokens":26123,"outputTokens":77,"cachedInputTokens":0,"cacheWriteTokens":0,"usageSource":"per_run","costStatus":"reported"}],"bundleSha256":"6dd8726e72545341f6772e54ec5aeb6264dec74db67290df44930d87135efb6b","toolCalls":[],"eventCounts":{"turn.started":1,"item.delta":1,"turn.completed":1},"eventTimes":{"session.started":"2026-10-09T19:19:01.000Z","turn.submitted":"2026-10-09T19:19:01.163Z","turn.started":"2026-10-09T19:19:01.893Z","firstTextDelta":"2026-10-09T19:19:07.169Z","turn.completed":"2026-10-09T19:19:07.384Z"},"completedRunDurationsMs":[15843.0]},{"index":5,"variant":"after","startedAt":"2026-10-09T19:19:10.050Z","measurementError":null,"ttfrMs":55217.199999809265,"domMs":55204.89999961853,"answer":"Hi 🙂","runCount":1,"statuses":["succeeded"],"usage":[{"model":"openrouter/deepseek/deepseek-v4-flash-0731","costUsd":0.00030421,"costUsdExact":"0.000304210","inputTokens":4294,"outputTokens":28,"cachedInputTokens":22543,"cacheWriteTokens":0,"usageSource":"per_run","costStatus":"reported"}],"bundleSha256":"6691475ed55f10179ec6ee0023a1f6462982f763848da7cd5b1d0baf38633b04","toolCalls":[],"eventCounts":{"turn.started":1,"item.delta":1,"turn.completed":1},"eventTimes":{"session.started":"2026-10-09T19:19:59.877Z","turn.submitted":"2026-10-09T19:19:59.990Z","turn.started":"2026-10-09T19:20:00.713Z","firstTextDelta":"2026-10-09T19:20:07.290Z","turn.completed":"2026-10-09T19:20:07.658Z"},"completedRunDurationsMs":[15104.0]},{"index":6,"variant":"after","startedAt":"2026-10-09T19:20:09.550Z","measurementError":null,"ttfrMs":63320.90000009537,"domMs":63313.09999990463,"answer":"Hi 👋","runCount":1,"statuses":["succeeded"],"usage":[{"model":"openrouter/deepseek/deepseek-v4-flash-0731","costUsd":0.00260723,"costUsdExact":"0.002607230","inputTokens":137971,"outputTokens":951,"cachedInputTokens":1024,"cacheWriteTokens":0,"usageSource":"per_run","costStatus":"reported"}],"bundleSha256":"6691475ed55f10179ec6ee0023a1f6462982f763848da7cd5b1d0baf38633b04","toolCalls":["paperclip_comment_on_task","paperclip_get_task_context","paperclip_comment_on_task","paperclip_finish"],"eventCounts":{"turn.started":1,"tool.execution.started":4,"item.delta":1,"turn.completed":1},"eventTimes":{"session.started":"2026-10-09T19:20:59.730Z","turn.submitted":"2026-10-09T19:20:59.859Z","turn.started":"2026-10-09T19:21:00.501Z","firstTextDelta":"2026-10-09T19:21:19.416Z","turn.completed":"2026-10-09T19:21:19.608Z"},"completedRunDurationsMs":[26519.0]},{"index":7,"variant":"before","startedAt":"2026-10-09T19:21:20.519Z","measurementError":null,"ttfrMs":47318.90000009537,"domMs":47246.800000190735,"answer":"Hi 👋","runCount":1,"statuses":["succeeded"],"usage":[{"model":"openrouter/deepseek/deepseek-v4-flash-0731","costUsd":0.000257965,"costUsdExact":"0.000257965","inputTokens":4700,"outputTokens":71,"cachedInputTokens":21407,"cacheWriteTokens":0,"usageSource":"per_run","costStatus":"reported"}],"bundleSha256":"6dd8726e72545341f6772e54ec5aeb6264dec74db67290df44930d87135efb6b","toolCalls":[],"eventCounts":{"turn.started":1,"item.delta":1,"turn.completed":1},"eventTimes":{"session.started":"2026-10-09T19:22:03.924Z","turn.submitted":"2026-10-09T19:22:04.439Z","turn.started":"2026-10-09T19:22:04.772Z","firstTextDelta":"2026-10-09T19:22:09.915Z","turn.completed":"2026-10-09T19:22:10.343Z"},"completedRunDurationsMs":[16702.0]},{"index":8,"variant":"before","startedAt":"2026-10-09T19:22:11.124Z","measurementError":null,"ttfrMs":48847.2000002861,"domMs":48802.59999990463,"answer":"Hi 🎉","runCount":1,"statuses":["succeeded"],"usage":[{"model":"openrouter/deepseek/deepseek-v4-flash-0731","costUsd":0.000327654,"costUsdExact":"0.000327654","inputTokens":44796,"outputTokens":32,"cachedInputTokens":0,"cacheWriteTokens":0,"usageSource":"per_run","costStatus":"reported"}],"bundleSha256":"6dd8726e72545341f6772e54ec5aeb6264dec74db67290df44930d87135efb6b","toolCalls":[],"eventCounts":{"turn.started":1,"item.delta":1,"turn.completed":1},"eventTimes":{"session.started":"2026-10-09T19:22:57.182Z","turn.submitted":"2026-10-09T19:22:57.281Z","turn.started":"2026-10-09T19:22:57.988Z","firstTextDelta":"2026-10-09T19:23:02.915Z","turn.completed":"2026-10-09T19:23:03.035Z"},"completedRunDurationsMs":[9541.0]},{"index":9,"variant":"after","startedAt":"2026-10-09T19:23:03.579Z","measurementError":null,"ttfrMs":29790.900000095367,"domMs":29748.800000190735,"answer":"Hi 👋","runCount":1,"statuses":["succeeded"],"usage":[{"model":"openrouter/deepseek/deepseek-v4-flash-0731","costUsd":0.00034054,"costUsdExact":"0.000340540","inputTokens":24710,"outputTokens":59,"cachedInputTokens":1792,"cacheWriteTokens":0,"usageSource":"per_run","costStatus":"reported"}],"bundleSha256":"6691475ed55f10179ec6ee0023a1f6462982f763848da7cd5b1d0baf38633b04","toolCalls":[],"eventCounts":{"turn.started":1,"item.delta":2,"turn.completed":1},"eventTimes":{"session.started":"2026-10-09T19:23:28.840Z","turn.submitted":"2026-10-09T19:23:28.954Z","turn.started":"2026-10-09T19:23:29.666Z","firstTextDelta":"2026-10-09T19:23:35.306Z","turn.completed":"2026-10-09T19:23:35.531Z"},"completedRunDurationsMs":[12732.0]},{"index":10,"variant":"after","startedAt":"2026-10-09T19:23:37.247Z","measurementError":null,"ttfrMs":61072.40000009537,"domMs":61061.300000190735,"answer":"Hi 👋","runCount":1,"statuses":["succeeded"],"usage":[{"model":"openrouter/deepseek/deepseek-v4-flash-0731","costUsd":0.00100571,"costUsdExact":"0.001005710","inputTokens":25307,"outputTokens":368,"cachedInputTokens":28160,"cacheWriteTokens":0,"usageSource":"per_run","costStatus":"reported"}],"bundleSha256":"6691475ed55f10179ec6ee0023a1f6462982f763848da7cd5b1d0baf38633b04","toolCalls":["paperclip_finish"],"eventCounts":{"turn.started":1,"tool.execution.started":1,"item.delta":1,"turn.completed":1},"eventTimes":{"session.started":"2026-10-09T19:24:33.721Z","turn.submitted":"2026-10-09T19:24:33.832Z","turn.started":"2026-10-09T19:24:34.601Z","firstTextDelta":"2026-10-09T19:24:43.135Z","turn.completed":"2026-10-09T19:24:43.278Z"},"completedRunDurationsMs":[17387.0]},{"index":11,"variant":"before","startedAt":"2026-10-09T19:24:44.710Z","measurementError":null,"ttfrMs":51311.7000002861,"domMs":51300.5,"answer":"Hi 👋","runCount":1,"statuses":["succeeded"],"usage":[{"model":"openrouter/deepseek/deepseek-v4-flash-0731","costUsd":0.00078825,"costUsdExact":"0.000788250","inputTokens":24316,"outputTokens":356,"cachedInputTokens":27648,"cacheWriteTokens":0,"usageSource":"per_run","costStatus":"reported"}],"bundleSha256":"6dd8726e72545341f6772e54ec5aeb6264dec74db67290df44930d87135efb6b","toolCalls":["paperclip_finish"],"eventCounts":{"turn.started":1,"tool.execution.started":1,"item.delta":1,"turn.completed":1},"eventTimes":{"session.started":"2026-10-09T19:25:31.347Z","turn.submitted":"2026-10-09T19:25:31.463Z","turn.started":"2026-10-09T19:25:32.208Z","firstTextDelta":"2026-10-09T19:25:39.405Z","turn.completed":"2026-10-09T19:25:39.563Z"},"completedRunDurationsMs":[15446.0]}],"additionalAttempts":[{"attempt":"setup-1","stage":"before_send","result":"non-editable composer locator","modelRequestObserved":false},{"attempt":"setup-2","stage":"before_send","result":"transpiler helper absent in browser","modelRequestObserved":false},{"attempt":"startup-pilot-before","runnerProfile":"debug","result":"no visible reply within 120s; OpenCode health startup SIGKILL; retained infrastructure failure","modelRequestObserved":false},{"attempt":"selector-pilot-after","runnerProfile":"debug","result":"Hi reply succeeded; browser selector missed it; no valid TTFR","providerReportedCostUsd":0.00085413}],"queueTimingCoverage":{"measuredTrials":10,"totalTrials":12,"records":[{"index":0,"runs":[{"createdAt":"2026-10-09T19:15:28.998Z","startedAt":"2026-10-09T19:15:53.558Z","queuedMs":24560.0}]},{"index":1,"runs":[{"createdAt":"2026-10-09T19:16:38.520Z","startedAt":"2026-10-09T19:16:54.157Z","queuedMs":15637.0}]},{"index":2,"runs":[{"createdAt":"2026-10-09T19:17:10.587Z","startedAt":"2026-10-09T19:17:23.561Z","queuedMs":12974.0}]},{"index":3,"runs":[{"createdAt":"2026-10-09T19:17:43.807Z","startedAt":"2026-10-09T19:17:53.564Z","queuedMs":9757.0}]},{"index":4,"runs":[{"createdAt":"2026-10-09T19:18:28.113Z","startedAt":"2026-10-09T19:18:54.011Z","queuedMs":25898.0}]},{"index":5,"runs":[{"createdAt":"2026-10-09T19:19:12.343Z","startedAt":"2026-10-09T19:19:53.993Z","queuedMs":41650.0}]},{"index":6,"runs":[{"createdAt":"2026-10-09T19:20:14.258Z","startedAt":"2026-10-09T19:20:53.866Z","queuedMs":39608.0}]},{"index":7,"runs":[{"createdAt":"2026-10-09T19:21:23.019Z","startedAt":"2026-10-09T19:21:54.298Z","queuedMs":31279.0}]},{"index":8,"runs":[{"createdAt":"2026-10-09T19:22:14.671Z","startedAt":"2026-10-09T19:22:53.933Z","queuedMs":39262.0}]},{"index":9,"runs":[{"createdAt":"2026-10-09T19:23:06.155Z","startedAt":"2026-10-09T19:23:23.941Z","queuedMs":17786.0}]}],"limitation":"The server shut down before a final API read; queue timing is retained for the first ten trials only."},"conclusion":"No reliable end-to-end speedup established. The after median is lower, but only three of six adjacent matched pairs were faster; median paired improvement is approximately zero. The deterministic streaming-forwarding fix remains separately supported.","limitations":["Local dev server and headless Chromium on macOS arm64; six fresh chats per variant; not a production/cloud latency guarantee.","Current product and current release runner held constant; only three OpenCode modules replaced by their exact pre-PR contents for baseline.","Default model reasoning unchanged; no no-reasoning override or decision policy.","Upstream routing/cache and queue time vary; trials run serially in predeclared counterbalanced ABBA order.","Render proxy: capture-phase Send click to visible nonempty Hi markdown, then two animation frames; screenshot verifies the rendered answer. This is not instrumented compositor pixel timing.","Two setup-only failures, one startup failure, and one invalid-measurement successful pilot are retained separately, not silently treated as successful timings."],"reproducibility":{"harnessSha256":"f93996b306c7d31c19bfcb9cc197653f867944ffb900d81a23a34f831e17c0c5","variantBuilderSha256":"4e88a7bbdc0de5d72601194ebe3b96d9c6f6eda5a33d8e0f00cf2d5d14db34c2","analyzerSha256":"2d129177a1b8033f768aad3c82dc8fff70ba0ba76e7270b2d8612ee760d4aaf4"}}
```
- Latest options-struct refactor: 33 provider tests and 86 backend tests
passed. Four port-dependent fixtures initially failed inside the sandbox
and passed with loopback access. Latest check snapshot: 51 successes, 4
skips, and 2 failures (`ci / e2e` and `ci / e2e shard (5/8)`). This
latency experiment does not make the PR merge-ready.
- Focused TypeScript suites passed: OpenCode driver, per-turn reasoning,
prepared-context recovery, proxy envelope, live sessions, Codex context
control, and native instruction measurements. The real runnerd/proxy
integration test confirms disabled followed by default in the same
session. A stale staged binary initially hid the new parameter;
rebuilding the release runner resolved it.
- Capability-guard regression suites passed all 53 tests; the real
runner/proxy boundary still accepts the per-turn selection. Both
OpenRouter and non-OpenRouter real runner/proxy cases passed, along with
62 existing OpenCode fixture tests. Runner TypeScript build passed after
the final correction.
- Affected Rust modules passed: 33 provider tests and 86 durable backend
tests. Malformed or unsupported reasoning selections do not dispatch a
provider turn or poison admission state.
- Pinned OpenCode 1.18.34 with a local OpenRouter-compatible fixture
sent three successive provider requests with reasoning omitted,
`enabled: false`, then omitted. Each request contained one copy of the
agent instructions. No paid provider call was needed.
- The reconstructed first-turn task envelope shrank from 4,356 to 460
characters. This is a payload-size measurement, not a claimed token or
latency reduction. Existing stored conversation history is not
rewritten.
- Prior head `8c07fe2e70267a7c8a3118f1071eca73ecdc6c8a`: all 53 checks
passed, with four intentional skips. Greptile is 5/5 with no open
findings. One bounded retry cleared first-test timeouts, a legacy
signoff lease failure, and CI runner shutdowns. The timed-out suites
also passed locally (14 auth-signal tests, 40 device-login tests, and
the runner resume case).
- Known reported diagnostic usage totals about $0.0029. Two prematurely
closed baseline warm-turn attempts are excluded from latency comparisons
and their usage is not fully captured; background provider usage is also
not fully accounted. No further paid runs are active.
- `pnpm -r typecheck`: passed. Runner TypeScript typechecks and build
also passed after the review correction.
- `pnpm test:run`: attempted and stopped after a reproducible failure in
the untouched Codex recovery test `archives a damaged prior epoch before
completing guarded same-task replacement with real runnerd`
(`server/src/services/native-runtime/native-session-resume.test.ts:1072`).
It returns `needs_review` / failed instead of done / succeeded,
including when rerun alone. The broad local suite is not green. The two
initial collection failures disappeared after building; their isolated
rerun passed all 1,565 tests.
- `pnpm build`: passed.
- Pinned real OpenCode plus a delayed local provider: first native delta
to Paperclip delta fell from 1,496 ms to less than 1 ms at millisecond
measurement resolution. This isolates forwarding delay, not
network/model inference.
- Two live driver samples after the patch reached first text in
2.28–2.62 seconds including session startup. Baseline first responses
took 12.37–13.43 seconds. Small unmatched samples had different
reasoning/cache outcomes, so these are observations, not an attributed
speedup ratio.
- Full compiled runnerd protocol diagnostic: completed with `Hi 👋`;
first assistant delta at 9.19 seconds and cleanup complete at 9.44
seconds. The seeded mock-control-plane harness used 27,512 input tokens
across two provider requests, split into 13,521 and 13,991. The extra
`answer_status_question` call posted `Hi 👋` to the mock task in 12 ms,
then caused another model request. A reconstruction contains 25 tools
(35.3k JSON characters), 12.8k system characters, and a 4.4k task
envelope; 3.9k instruction characters appear in both system and task
context. Provider-reported cost was $0.00073976; the pricing-table
estimate was $0.00395332. This is one integration diagnostic, not a
production latency claim.
- A separate full-runner sample with existing lazy tool discovery also
completed, but took 14.64 seconds through cleanup and called
`paperclip_finish`. Its usage ledger recorded 12,888 input and 11,776
cached-input tokens, with $0.00096344 provider-reported cost. This did
not establish a latency improvement, so the default tool exposure was
not changed.
- Four direct OpenRouter requests separately measured default reasoning
at 2.45/3.90 seconds to first text and disabled reasoning at 0.66/0.53
seconds. Default reasoning remains unchanged; disabling it requires the
explicit per-turn option.
## Risks
- Immediate readiness dispatch changes scheduling timing. It uses
existing admission checks and preserves the periodic sweep as recovery.
- Native model verification relies on the real request rather than
catalog enumeration. The final live batch had provider startup failures;
overall speed and reliability are not qualified.
- Incremental chunks without IDs cannot safely be deduplicated by
content. Repeated identical chunks are valid. Known-ID duplicates remain
suppressed.
- OpenCode uses its bundled model catalog plus the explicit
selected-model entry. New remote catalog metadata will not refresh
during isolated startup.
- Default provider plugins are disabled. This runner supplies
credentials and assigned MCP tools through its own controlled
configuration.
- Disabled reasoning is supported only on the OpenCode/OpenRouter path
and requires a compatible model. It applies to one turn and is not an
automatic latency policy.
- The matched live browser sample is small and does not establish a
reliable end-to-end improvement. It includes local queueing/startup and
browser rendering, but does not qualify production or remote sandbox
latency.
## Model Used
OpenAI GPT-6 (Codex). Tool use, code editing, and command execution were
used. The session does not expose a more specific model ID or
context-window size. Live diagnostic target: DeepSeek
`deepseek/deepseek-v4-flash-0731` through OpenRouter and OpenCode
1.18.34.
## 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
- [ ] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] All Paperclip CI gates are green
- [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [x] I will address all Greptile and reviewer comments before
requesting merge
---------
Co-authored-by: Paperclip
---
doc/AGENT-LIFECYCLE.md | 13 +++
.../src/server/test.remote.test.ts | 59 +++++++++++-
.../opencode-local/src/server/test.ts | 34 ++++++-
packages/paperclip-runner/README.md | 31 ++++++
.../crates/runner-core/src/codex_provider.rs | 27 +++++-
.../runner-core/src/provider_backend.rs | 49 +++++++++-
.../backends/harness-driver-backend.test.ts | 28 ++++++
.../src/backends/harness-driver-backend.ts | 4 +
.../native-instruction-measurement.test.ts | 3 +-
.../src/cli/opencode-app-server-proxy.test.ts | 42 +++++----
.../src/cli/opencode-app-server-proxy.ts | 2 +
.../src/cli/opencode-proxy-command.ts | 11 ++-
.../src/cli/opencode-proxy-task-envelope.ts | 3 +-
.../src/contracts/harness-driver.ts | 3 +
.../src/contracts/native-session-backend.ts | 2 +
.../paperclip-runner/src/contracts/types.ts | 2 +
.../src/drivers/codex/app-server-transport.ts | 1 +
...-app-server-driver.context-control.test.ts | 13 +++
.../drivers/codex/codex-harness-session.ts | 10 ++
.../opencode/opencode-server-driver.test.ts | 75 ++++++++++++++-
.../opencode/opencode-server-driver.ts | 94 +++++++++++++++----
.../prepared-context-transport.test.ts | 4 +
.../drivers/opencode/reasoning-mode.test.ts | 63 +++++++++++++
.../src/drivers/opencode/reasoning-mode.ts | 8 ++
.../paperclip-runner/src/live/live-session.ts | 3 +-
.../src/live/runnerd-codex-transport.test.ts | 23 ++++-
.../src/live/runnerd-codex-transport.ts | 20 ++++
.../test/fixtures/fake-opencode-server.mjs | 50 +++++++++-
...artbeat-responsible-user-invariant.test.ts | 22 +++++
.../__tests__/diagnostics.test.ts | 48 +++++++++-
.../agent-lifecycle/application/ports.ts | 4 +-
.../agent-lifecycle/application/worker.ts | 8 ++
server/src/services/agent-lifecycle-driver.ts | 3 +
server/src/services/heartbeat.ts | 6 ++
34 files changed, 707 insertions(+), 61 deletions(-)
create mode 100644 packages/paperclip-runner/src/drivers/opencode/reasoning-mode.test.ts
create mode 100644 packages/paperclip-runner/src/drivers/opencode/reasoning-mode.ts
diff --git a/doc/AGENT-LIFECYCLE.md b/doc/AGENT-LIFECYCLE.md
index 96767c83131..87e0967316f 100644
--- a/doc/AGENT-LIFECYCLE.md
+++ b/doc/AGENT-LIFECYCLE.md
@@ -54,6 +54,19 @@ An old result cannot complete a newer operation.
Deletion requires completed termination or a rejected hire.
The existing restrictions for built-in agents and accounting still apply.
+Messages accepted during setup remain queued until verification completes.
+After the worker commits `ready`, its driver immediately asks the scheduler to
+dispatch that agent's saved work. The scheduler rechecks company scope,
+invokability, budgets, concurrency, and task ownership through normal admission.
+If dispatch fails, readiness remains committed and the periodic queue sweep
+retries; setup is not repeated because of a scheduler failure.
+
+Native OpenCode verification checks the selected model with the real hello
+probe. It validates the model identifier but does not first enumerate the full model
+catalog. Authentication failures, unavailable models, and probe timeouts still
+prevent readiness. Without an explicit local working directory, the native
+probe uses a temporary empty directory instead of the server's checkout.
+
## Commands and transactions
`requestHire` writes the agent record before preparation starts.
diff --git a/packages/adapters/opencode-local/src/server/test.remote.test.ts b/packages/adapters/opencode-local/src/server/test.remote.test.ts
index aebe6cfa0a9..b9a63a48b97 100644
--- a/packages/adapters/opencode-local/src/server/test.remote.test.ts
+++ b/packages/adapters/opencode-local/src/server/test.remote.test.ts
@@ -1,4 +1,4 @@
-import { mkdtemp, rm } from "node:fs/promises";
+import { mkdtemp, rm, stat } from "node:fs/promises";
import os from "node:os";
import path from "node:path";
import { afterEach, beforeEach, describe, expect, it, vi } from "vitest";
@@ -72,6 +72,13 @@ vi.mock("@paperclipai/adapter-utils/execution-target", async () => {
});
import { testEnvironment } from "./test.js";
+import { discoverOpenCodeModels, ensureOpenCodeModelConfiguredAndAvailable } from "./models.js";
+
+vi.mock("./models.js", async () => ({
+ ...await vi.importActual("./models.js"),
+ discoverOpenCodeModels: vi.fn().mockRejectedValue(new Error("catalog should not be needed")),
+ ensureOpenCodeModelConfiguredAndAvailable: vi.fn().mockRejectedValue(new Error("catalog should not be needed")),
+}));
describe("opencode remote environment diagnostics", () => {
const configHomes: string[] = [];
@@ -89,6 +96,56 @@ describe("opencode remote environment diagnostics", () => {
await Promise.all(configHomes.splice(0).map(dir => rm(dir, { recursive: true, force: true })));
});
+ it.each(["passed", "auth_required", "model_unavailable", "timed_out"])(
+ "validates a native model with the real probe path and no catalog scan (%s)", async outcome => {
+ const successfulProbe = await runAdapterExecutionTargetProcess.getMockImplementation()!();
+ runAdapterExecutionTargetProcess.mockResolvedValueOnce(outcome === "passed" ? successfulProbe : {
+ ...successfulProbe, exitCode: 1, stdout: "", timedOut: outcome === "timed_out",
+ stderr: outcome === "auth_required" ? "invalid API key" : "ProviderModelNotFoundError",
+ });
+ const result = await testEnvironment({ companyId: "company-1", adapterType: "paperclip_runner",
+ config: { model: "openrouter/deepseek/deepseek-v4-flash-0731", env: { XDG_CONFIG_HOME: configHome } } });
+ expect(discoverOpenCodeModels).not.toHaveBeenCalled();
+ expect(ensureOpenCodeModelConfiguredAndAvailable).not.toHaveBeenCalled();
+ expect(runAdapterExecutionTargetProcess).toHaveBeenCalledExactlyOnceWith(
+ expect.any(String), null, "opencode", ["run", "--format", "json", "--model", "openrouter/deepseek/deepseek-v4-flash-0731"],
+ expect.objectContaining({ stdin: "Respond with hello." }),
+ );
+ expect(result.status).toBe(outcome === "passed" ? "pass" : "warn");
+ expect(result.checks).toContainEqual(expect.objectContaining({ code: `opencode_hello_probe_${outcome}` }));
+ const call = runAdapterExecutionTargetProcess.mock.calls[0] as unknown as [unknown, unknown, unknown, unknown, { cwd: string }];
+ expect(path.basename(call[4].cwd)).toMatch(/^paperclip-opencode-native-probe-/);
+ await expect(stat(call[4].cwd)).rejects.toMatchObject({ code: "ENOENT" });
+ },
+ );
+
+ it("rejects a malformed native model before starting the probe", async () => {
+ const result = await testEnvironment({ companyId: "company-1", adapterType: "paperclip_runner",
+ config: { model: "missing-provider", env: { XDG_CONFIG_HOME: configHome } } });
+ expect(result.status).toBe("fail");
+ expect(runAdapterExecutionTargetProcess).not.toHaveBeenCalled();
+ expect(discoverOpenCodeModels).not.toHaveBeenCalled();
+ });
+
+ it("preserves an explicitly configured native probe directory", async () => {
+ await testEnvironment({ companyId: "company-1", adapterType: "paperclip_runner",
+ config: { model: "provider/model", cwd: configHome, env: { XDG_CONFIG_HOME: configHome } } });
+ const call = runAdapterExecutionTargetProcess.mock.calls[0] as unknown as [unknown, unknown, unknown, unknown, { cwd: string }];
+ expect(call[4].cwd).toBe(configHome);
+ expect((await stat(configHome)).isDirectory()).toBe(true);
+ });
+
+ it("keeps catalog validation for the legacy local adapter", async () => {
+ vi.mocked(discoverOpenCodeModels).mockResolvedValueOnce([{ id: "provider/model", label: "Model" }]);
+ vi.mocked(ensureOpenCodeModelConfiguredAndAvailable).mockResolvedValueOnce([{ id: "provider/model", label: "Model" }]);
+ const result = await testEnvironment({ companyId: "company-1", adapterType: "opencode_local",
+ config: { model: "provider/model", env: { XDG_CONFIG_HOME: configHome } } });
+ expect(result.status).toBe("pass");
+ expect(discoverOpenCodeModels).toHaveBeenCalledOnce();
+ expect(ensureOpenCodeModelConfiguredAndAvailable).toHaveBeenCalledOnce();
+ expect(runAdapterExecutionTargetProcess).toHaveBeenCalledOnce();
+ });
+
it.each([false, true])("stages remote runtime config assets for sandbox hello probes (managed=%s)", async (managed) => {
const configHome = await mkdtemp(path.join(os.tmpdir(), "opencode-remote-test-config-"));
configHomes.push(configHome);
diff --git a/packages/adapters/opencode-local/src/server/test.ts b/packages/adapters/opencode-local/src/server/test.ts
index 7f2480b9c90..628745e9772 100644
--- a/packages/adapters/opencode-local/src/server/test.ts
+++ b/packages/adapters/opencode-local/src/server/test.ts
@@ -25,7 +25,7 @@ import {
prepareAdapterExecutionTargetRuntime,
overrideAdapterExecutionTargetRemoteCwd,
} from "@paperclipai/adapter-utils/execution-target";
-import { discoverOpenCodeModels, ensureOpenCodeModelConfiguredAndAvailable } from "./models.js";
+import { discoverOpenCodeModels, ensureOpenCodeModelConfiguredAndAvailable, requireOpenCodeModelId } from "./models.js";
import { parseOpenCodeJsonl } from "./parse.js";
import { SANDBOX_INSTALL_COMMAND } from "../index.js";
import { prepareOpenCodeRuntimeConfig, prepareManagedOpenCodeRemoteHomes } from "./runtime-config.js";
@@ -141,9 +141,16 @@ export async function testEnvironment(
// still has the path available for cleanup in `finally` — otherwise the
// `fs.mkdtemp` directory leaks on the early-throw path.
let preparedRuntimeWorkspaceLocalDir: string | null = null;
+ let nativeProbeWorkspaceLocalDir: string | null = null;
try {
let runtimeTarget: AdapterExecutionTarget | null = target ?? null;
let runtimeCwd = cwd;
+ // Native workspaces are assigned at execution time. A configuration probe
+ // without an explicit workspace must not run in the server's checkout.
+ if (!targetIsRemote && ctx.adapterType === "paperclip_runner" && !asString(config.cwd, "").trim()) {
+ nativeProbeWorkspaceLocalDir = await fs.mkdtemp(path.join(os.tmpdir(), "paperclip-opencode-native-probe-"));
+ runtimeCwd = nativeProbeWorkspaceLocalDir;
+ }
if (targetIsRemote) {
preparedRuntimeWorkspaceLocalDir = await fs.mkdtemp(path.join(os.tmpdir(), `paperclip-opencode-envtest-${runId}-`));
const preparedExecutionTargetRuntime = await prepareAdapterExecutionTargetRuntime({
@@ -223,12 +230,30 @@ export async function testEnvironment(
let modelValidationPassed = false;
const configuredModel = asString(config.model, "").trim();
+ // The native runner already selects a concrete model. Its real hello probe
+ // verifies model access and credentials; a full catalog scan adds another
+ // cold CLI launch (and discovery retries) without verifying that request.
+ const nativeModelProbe = ctx.adapterType === "paperclip_runner" && Boolean(configuredModel);
+
// Model discovery and validation use local child processes against
// OpenCode's `models` subcommand and JSON config; these are not yet
// wired through the execution target. When probing a remote env, skip
// discovery/validation and rely on the remote hello probe to surface
// model/auth issues directly.
- if (targetIsRemote && configuredModel) {
+ if (nativeModelProbe) {
+ try {
+ requireOpenCodeModelId(configuredModel);
+ modelValidationPassed = true;
+ checks.push({
+ code: "opencode_model_validation_via_probe",
+ level: "info",
+ message: "The configured native-runner model will be validated by the hello probe.",
+ });
+ } catch (error) {
+ checks.push({ code: "opencode_model_invalid", level: "error",
+ message: error instanceof Error ? error.message : "Configured model is invalid." });
+ }
+ } else if (targetIsRemote && configuredModel) {
checks.push({
code: "opencode_model_validation_skipped_remote",
level: "info",
@@ -305,7 +330,7 @@ export async function testEnvironment(
const modelUnavailable = checks.some((check) => check.code === "opencode_hello_probe_model_unavailable");
if (!configuredModel && !modelUnavailable) {
// No model configured – skip model requirement if no model-related checks exist
- } else if (!targetIsRemote && configuredModel && canRunProbe) {
+ } else if (!nativeModelProbe && !targetIsRemote && configuredModel && canRunProbe) {
try {
await ensureOpenCodeModelConfiguredAndAvailable({
model: configuredModel,
@@ -430,6 +455,9 @@ export async function testEnvironment(
}
}
} finally {
+ if (nativeProbeWorkspaceLocalDir) {
+ await fs.rm(nativeProbeWorkspaceLocalDir, { recursive: true, force: true }).catch(() => {});
+ }
await restoreWorkspace?.();
if (!restoreWorkspace && preparedRuntimeWorkspaceLocalDir) {
// Reached when `prepareAdapterExecutionTargetRuntime` threw before
diff --git a/packages/paperclip-runner/README.md b/packages/paperclip-runner/README.md
index f87a1d07652..a7f0647a39f 100644
--- a/packages/paperclip-runner/README.md
+++ b/packages/paperclip-runner/README.md
@@ -238,6 +238,37 @@ provider-native operations. Full auto is resolved consistently for agent
creation, adapter conversion, direct driver launches, and fresh/resumed turns.
Explicitly stored restrictive modes still apply.
+Native OpenCode forwards incremental text and reasoning parts as they arrive;
+the final full snapshot selects the final response without repeating streamed
+text. Delta parts must match an observed message and part identity. Identical
+chunks without event IDs remain distinct tokens. Each isolated launch disables
+automatic updates, remote model-catalog refreshes, and default plugins; the
+runner supplies the pinned executable, selected model, and assigned MCP tools.
+Provider inference and reasoning still contribute to time to first text. These
+startup settings do not disable model reasoning.
+
+On macOS, the verified OpenCode executable keeps its private launch pathname
+until the provider process exits. Removing that pathname immediately after
+spawn can kill or stall the signed binary before its health check succeeds.
+Ownership, permissions, and file-identity checks remain enforced; cleanup also
+runs on failed launches, and retries create a fresh launch snapshot.
+
+Call `session.startTurn({ message, reasoningMode: "disabled" })` to disable
+reasoning for one OpenCode/OpenRouter turn. `CapabilityLiveSession.sendMessage`
+accepts the same option in its second argument. The choice travels in the durable
+`turn.start` command and selects an OpenCode model variant for that prompt only.
+Omitting it (or passing `"default"`) on the next turn restores provider defaults,
+including on a warm or recovered session. It is not an environment variable or
+agent-wide setting. No automatic decision policy selects it. The model must
+support disabling reasoning; other runner providers reject an explicit selection
+before starting work. Native sessions expose support through the
+`perTurnReasoning` capability.
+
+OpenCode's proxy sends agent instructions once per turn through the system prompt. The
+task envelope retains task-specific constraints and the completion contract,
+without repeating the system instructions. This applies to all users of the
+proxy; already-recorded conversation history is not rewritten.
+
The runner's authenticated bridge and controller still enforce company access,
action claims, task modes, and governed approvals. Provider permission defaults
do not change workspace isolation or grant credentials or connection access.
diff --git a/packages/paperclip-runner/runner/crates/runner-core/src/codex_provider.rs b/packages/paperclip-runner/runner/crates/runner-core/src/codex_provider.rs
index 29494a4e21a..61432b9dbe0 100644
--- a/packages/paperclip-runner/runner/crates/runner-core/src/codex_provider.rs
+++ b/packages/paperclip-runner/runner/crates/runner-core/src/codex_provider.rs
@@ -386,6 +386,13 @@ pub struct CodexSkillInput {
pub path: String,
}
+/// Optional inputs for one turn; omitted fields preserve provider defaults.
+#[derive(Clone, Copy, Debug, Default)]
+pub struct CodexTurnOptions<'a> {
+ pub skills: &'a [CodexSkillInput],
+ pub reasoning_mode: Option<&'a str>,
+}
+
impl CodexSkillInput {
pub fn validate(&self) -> Result<(), LocalRunnerError> {
if self.input_type != "skill"
@@ -1670,15 +1677,26 @@ impl CodexProvider {
}
pub fn start_turn(&mut self, message: &str, cwd: &str) -> Result {
- self.start_turn_with_skills(message, cwd, &[])
+ self.start_turn_with_options(message, cwd, CodexTurnOptions::default())
}
- pub fn start_turn_with_skills(
+ pub fn start_turn_with_options(
&mut self,
message: &str,
cwd: &str,
- skills: &[CodexSkillInput],
+ options: CodexTurnOptions<'_>,
) -> Result {
+ let CodexTurnOptions {
+ skills,
+ reasoning_mode,
+ } = options;
+ if reasoning_mode.is_some_and(|mode| {
+ self.config.provider != "opencode" || !matches!(mode, "default" | "disabled")
+ }) {
+ return Err(LocalRunnerError::invalid(
+ "turn.reasoningMode requires OpenCode and must be default or disabled",
+ ));
+ }
if skills.len() > 64 || (self.config.provider != "codex" && !skills.is_empty()) {
return Err(LocalRunnerError::invalid(
"explicit skills require Codex and at most 64 selections",
@@ -1727,6 +1745,9 @@ impl CodexProvider {
let turn_params_object = turn_params
.as_object_mut()
.expect("Codex turn parameters are an object");
+ if let Some(mode) = reasoning_mode {
+ turn_params_object.insert("reasoningMode".to_owned(), json!(mode));
+ }
if self.permission_profile == "paperclip-runner-external-sandbox" {
turn_params_object.insert(
"sandboxPolicy".to_owned(),
diff --git a/packages/paperclip-runner/runner/crates/runner-core/src/provider_backend.rs b/packages/paperclip-runner/runner/crates/runner-core/src/provider_backend.rs
index 23268296bd6..329c3dc7bf4 100644
--- a/packages/paperclip-runner/runner/crates/runner-core/src/provider_backend.rs
+++ b/packages/paperclip-runner/runner/crates/runner-core/src/provider_backend.rs
@@ -13,7 +13,7 @@ use serde_json::{json, Value};
use sha2::{Digest, Sha256};
use crate::codex_provider::{
- CodexProvider, CodexProviderConfig, CodexProviderEvent, CodexSkillInput,
+ CodexProvider, CodexProviderConfig, CodexProviderEvent, CodexSkillInput, CodexTurnOptions,
ProviderStartupObservation, ProviderStartupStage, RejectedAcceptedTurn,
MAX_SETTLED_PROVIDER_TURN_IDS,
};
@@ -2903,7 +2903,24 @@ impl CodexCommandExecutor {
.validate()
.map_err(|error| DurableRunnerError::invalid(error.to_string()))?;
}
+ let reasoning_mode = match payload.get("reasoningMode") {
+ None => None,
+ Some(Value::String(mode)) if matches!(mode.as_str(), "default" | "disabled") => {
+ Some(mode.as_str())
+ }
+ Some(_) => return Err(DurableRunnerError::invalid("invalid turn.reasoningMode")),
+ };
self.restore_provider_if_needed()?;
+ if reasoning_mode.is_some()
+ && self
+ .state
+ .as_ref()
+ .is_none_or(|state| state.config.provider != "opencode")
+ {
+ return Err(DurableRunnerError::invalid(
+ "turn.reasoningMode requires OpenCode",
+ ));
+ }
if self
.state
.as_ref()
@@ -2986,7 +3003,14 @@ impl CodexCommandExecutor {
rejected_accepted_turn,
) = {
let provider = self.ensure_provider()?;
- let result = provider.start_turn_with_skills(text, &cwd, &skills);
+ let result = provider.start_turn_with_options(
+ text,
+ &cwd,
+ CodexTurnOptions {
+ skills: &skills,
+ reasoning_mode,
+ },
+ );
(
result,
provider.completed_turn_authority().is_some(),
@@ -6068,6 +6092,27 @@ mod tests {
executor.state = Some(state);
executor.restore_checked = true;
+ for mode in [
+ json!(null),
+ json!(false),
+ json!("bad"),
+ json!("disabled"),
+ json!("default"),
+ ] {
+ let error = executor
+ .start_turn(&json!({"text": "must not reach the provider", "reasoningMode": mode}))
+ .unwrap_err();
+ assert!(error.to_string().contains("reasoningMode"));
+ assert!(executor.provider.is_none());
+ assert!(
+ !executor
+ .state
+ .as_ref()
+ .unwrap()
+ .ambiguous_turn_start_pending
+ );
+ }
+
let error = executor
.start_turn(&json!({"text": "must not reach the provider"}))
.expect_err("the exact regular backlog limit must reject admission");
diff --git a/packages/paperclip-runner/src/backends/harness-driver-backend.test.ts b/packages/paperclip-runner/src/backends/harness-driver-backend.test.ts
index e1542606a24..a5cc423e60e 100644
--- a/packages/paperclip-runner/src/backends/harness-driver-backend.test.ts
+++ b/packages/paperclip-runner/src/backends/harness-driver-backend.test.ts
@@ -97,6 +97,34 @@ const driver: HarnessDriver = {
};
describe("HarnessDriverBackend", () => {
+ it.each(["default", "disabled"] as const)("rejects unsupported reasoning %s before starting provider work", async reasoningMode => {
+ const startTurn = vi.fn(async () => ({ turnId: "turn-1" }));
+ const session = Object.assign(new FakeHarnessSession(), { startTurn });
+ const backend = new HarnessDriverBackend({ ...driver, openSession: async () => session });
+ const opened = await backend.openSession({ identity: {
+ runId: "run-1", sessionId: "session-1", companyId: "company-1", issueId: "issue-1", agentId: "agent-1",
+ } });
+ expect((await opened.capabilities()).perTurnReasoning).toBe(false);
+ const message = { role: "user" as const, text: "Hi" };
+ await expect(opened.startTurn({ message, reasoningMode })).rejects.toThrow("Per-turn reasoning is not supported");
+ expect(startTurn).not.toHaveBeenCalled();
+ await expect(opened.startTurn({ message })).resolves.toEqual({ turnId: "turn-1" });
+ expect(startTurn).toHaveBeenCalledExactlyOnceWith({ message });
+ });
+
+ it("forwards reasoning when the provider explicitly supports it", async () => {
+ const startTurn = vi.fn(async () => ({ turnId: "turn-1" }));
+ const session = Object.assign(new FakeHarnessSession(), { startTurn, supportsTurnReasoning: () => true });
+ const backend = new HarnessDriverBackend({ ...driver, openSession: async () => session });
+ const opened = await backend.openSession({ identity: {
+ runId: "run-1", sessionId: "session-1", companyId: "company-1", issueId: "issue-1", agentId: "agent-1",
+ } });
+ expect((await opened.capabilities()).perTurnReasoning).toBe(true);
+ const input = { message: { role: "user" as const, text: "Hi" }, reasoningMode: "disabled" as const };
+ await opened.startTurn(input);
+ expect(startTurn).toHaveBeenCalledExactlyOnceWith(input);
+ });
+
it("refreshes negotiated controls after a warm handshake and separates follow-ups", async () => {
let controls = { steering: false, queuedFollowUp: false };
const steer = vi.fn(async () => {});
diff --git a/packages/paperclip-runner/src/backends/harness-driver-backend.ts b/packages/paperclip-runner/src/backends/harness-driver-backend.ts
index 6e3897e4e57..fcdb43e16b7 100644
--- a/packages/paperclip-runner/src/backends/harness-driver-backend.ts
+++ b/packages/paperclip-runner/src/backends/harness-driver-backend.ts
@@ -454,6 +454,7 @@ class HarnessNativeSession implements NativeSession {
return {
resume: true,
typedEvents: true,
+ perTurnReasoning: this.#session.supportsTurnReasoning?.() === true,
steering: (negotiated?.steering ?? this.#steeringSupported) && this.#session.steer !== undefined,
queuedFollowUp: negotiated?.queuedFollowUp === true && this.#session.steer !== undefined,
interruption: this.#session.interrupt !== undefined,
@@ -710,6 +711,9 @@ class HarnessNativeSession implements NativeSession {
// The provider has no active turn to interrupt yet. Do not launch work
// whose events cancellation would suppress and leave the owner waiting.
if (this.#explicitlyCancelled) throw new Error("native_session_cancelled");
+ if (input.reasoningMode !== undefined && this.#session.supportsTurnReasoning?.() !== true) {
+ throw new Error("Per-turn reasoning is not supported by this provider");
+ }
this.#terminal = null;
try {
const started = await this.#session.startTurn(input);
diff --git a/packages/paperclip-runner/src/backends/native-instruction-measurement.test.ts b/packages/paperclip-runner/src/backends/native-instruction-measurement.test.ts
index 98507b24338..2c5592af47e 100644
--- a/packages/paperclip-runner/src/backends/native-instruction-measurement.test.ts
+++ b/packages/paperclip-runner/src/backends/native-instruction-measurement.test.ts
@@ -190,11 +190,10 @@ describe("direct OpenCode HTTP instruction boundary", () => {
const request = requests.at(-1)!;
expect(request.providerID).toBe("openrouter"); expect(request.modelID).toBe("deepseek/deepseek-v4-flash-0731");
const text = request.parts[0].text as string;
+ expect(request.system).toBe(nativeSystemInstructions(value));
if (schema === "v4" && phase === "start") {
- expect(request.system).toBe(nativeSystemInstructions(input));
expect(JSON.parse(text).task.constraints).toEqual(nativeTaskConstraints(input));
} else {
- expect(request.system).toBeUndefined();
expect(JSON.parse(text)).toEqual(envelope);
}
directOpenCodeReceipts.push({ provider: "opencode", schema, phase, request: measure(request),
diff --git a/packages/paperclip-runner/src/cli/opencode-app-server-proxy.test.ts b/packages/paperclip-runner/src/cli/opencode-app-server-proxy.test.ts
index 0c672b356af..65250a0f8d5 100644
--- a/packages/paperclip-runner/src/cli/opencode-app-server-proxy.test.ts
+++ b/packages/paperclip-runner/src/cli/opencode-app-server-proxy.test.ts
@@ -9,7 +9,7 @@ import {
rmSync,
writeFileSync,
} from "node:fs";
-import { spawnSync } from "node:child_process";
+import { spawn, spawnSync } from "node:child_process";
import { tmpdir } from "node:os";
import { join } from "node:path";
import {
@@ -28,7 +28,7 @@ describe("OpenCode runnerd proxy executable", () => {
])).toEqual({ command: "/proc/self/fd/7", commandFd: 7 });
});
- it("executes the inherited artifact through the nested child fd mapping", () => {
+ it("executes the inherited artifact through the nested child fd mapping", async () => {
if (process.platform === "win32") return;
if (process.platform === "darwin") {
const root = mkdtempSync(join(tmpdir(), "paperclip-opencode-nested-"));
@@ -45,22 +45,27 @@ describe("OpenCode runnerd proxy executable", () => {
TRUSTED_OPENCODE_EXECUTABLE_ARG,
command,
]);
- binding.commandLifecycle?.beforeSpawn();
- const child = spawnSync(binding.command, [], { encoding: "utf8" });
- expect(child.error).toBeUndefined();
- expect(child.status).toBe(0);
- expect(child.stdout).toBe("verified-nested-spawn");
- binding.commandLifecycle?.afterSpawn();
- expect(() => lstatSync(command)).toThrow();
- expect(() => lstatSync(directory)).toThrow();
- binding.commandLifecycle?.beforeSpawn();
- const retriedChild = spawnSync(binding.command, [], { encoding: "utf8" });
- expect(retriedChild.error).toBeUndefined();
- expect(retriedChild.status).toBe(0);
- expect(retriedChild.stdout).toBe("verified-nested-spawn");
- binding.commandLifecycle?.afterSpawn();
- expect(() => lstatSync(command)).toThrow();
- expect(() => lstatSync(directory)).toThrow();
+ for (let attempt = 0; attempt < 2; attempt += 1) {
+ binding.commandLifecycle!.beforeSpawn();
+ const child = spawn(binding.command, [], { stdio: ["ignore", "pipe", "pipe"] });
+ let stdout = "";
+ child.stdout!.on("data", chunk => { stdout += chunk; });
+ const exited = new Promise((resolve, reject) => {
+ child.once("error", reject);
+ child.once("close", resolve);
+ });
+ binding.commandLifecycle!.afterSpawn();
+ // Exercise the production async boundary, not spawnSync followed by
+ // cleanup after the executable has already finished.
+ expect(lstatSync(command).isFile()).toBe(true);
+ expect(() => binding.commandLifecycle!.beforeSpawn()).toThrow("still in use");
+ expect(await exited).toBe(0);
+ expect(stdout).toBe("verified-nested-spawn");
+ binding.commandLifecycle!.afterExit();
+ binding.commandLifecycle!.afterExit();
+ expect(() => lstatSync(command)).toThrow();
+ expect(() => lstatSync(directory)).toThrow();
+ }
} finally {
rmSync(root, { recursive: true, force: true });
}
@@ -131,6 +136,7 @@ describe("OpenCode runnerd proxy task envelope", () => {
criterionIds: ["objective", "verification"],
},
})).toMatchObject({
+ constraints: ["Use Paperclip MCP tools for semantic operations."],
completionContract: {
revision: "17",
criteria: [
diff --git a/packages/paperclip-runner/src/cli/opencode-app-server-proxy.ts b/packages/paperclip-runner/src/cli/opencode-app-server-proxy.ts
index 11cc112dea7..0e60ac1f245 100644
--- a/packages/paperclip-runner/src/cli/opencode-app-server-proxy.ts
+++ b/packages/paperclip-runner/src/cli/opencode-app-server-proxy.ts
@@ -1,4 +1,5 @@
#!/usr/bin/env node
+import { parseOpenCodeReasoningMode } from "../drivers/opencode/reasoning-mode.js";
import { createInterface } from "node:readline";
import { readFileSync } from "node:fs";
import { resolve } from "node:path";
@@ -365,6 +366,7 @@ async function handle(message: RpcMessage): Promise {
.join("\n");
const turn = await session.startTurn({
message: { role: "user", text: messageText },
+ ...(params.reasoningMode === undefined ? {} : { reasoningMode: parseOpenCodeReasoningMode(params.reasoningMode) }),
});
activeTurnId = turn.turnId;
// OpenCode normally publishes session/turn startup over SSE, but a fast
diff --git a/packages/paperclip-runner/src/cli/opencode-proxy-command.ts b/packages/paperclip-runner/src/cli/opencode-proxy-command.ts
index 2da3349bf34..3d73ce8d22b 100644
--- a/packages/paperclip-runner/src/cli/opencode-proxy-command.ts
+++ b/packages/paperclip-runner/src/cli/opencode-proxy-command.ts
@@ -39,6 +39,7 @@ export function trustedOpenCodeLaunchBinding(
commandLifecycle?: {
beforeSpawn(): void;
afterSpawn(): void;
+ afterExit(): void;
};
} {
const command = args.length === 2 && args[0] === TRUSTED_OPENCODE_EXECUTABLE_ARG
@@ -119,8 +120,7 @@ export function trustedOpenCodeLaunchBinding(
let materializedIdentity: { dev: number; ino: number } | undefined;
const materializeForSpawn = () => {
if (materialized) {
- validateMacSnapshot(materializedIdentity);
- return;
+ throw new Error("Verified OpenCode executable is still in use");
}
let destinationFd: number | undefined;
try {
@@ -184,10 +184,15 @@ export function trustedOpenCodeLaunchBinding(
// attackers are outside the documented local-host trust boundary in
// docs/durable-recovery.md. Keep the verified source on an unlinked
// descriptor, rematerialize only at the syscall boundary (including
- // retries), then remove the executable pathname immediately.
+ // retries). Keep its private pathname until the child exits: deleting
+ // a signed executable during macOS startup can kill or stall the child.
beforeSpawn: materializeForSpawn,
afterSpawn() {
validateMacSnapshot(materializedIdentity);
+ },
+ afterExit() {
+ if (!materialized) return;
+ validateMacSnapshot(materializedIdentity);
unlinkSync(command);
rmdirSync(dirname(command));
materialized = false;
diff --git a/packages/paperclip-runner/src/cli/opencode-proxy-task-envelope.ts b/packages/paperclip-runner/src/cli/opencode-proxy-task-envelope.ts
index dd30c666d90..535b64d6d30 100644
--- a/packages/paperclip-runner/src/cli/opencode-proxy-task-envelope.ts
+++ b/packages/paperclip-runner/src/cli/opencode-proxy-task-envelope.ts
@@ -30,7 +30,8 @@ export function openCodeProxyTaskEnvelope(params: Record) {
}
: {}),
constraints: [
- text(params.baseInstructions, "Complete only the supplied task."),
+ // baseInstructions already travels in the system prompt. Keep task
+ // constraints here without duplicating the agent instruction bundle.
"Use Paperclip MCP tools for semantic operations.",
],
});
diff --git a/packages/paperclip-runner/src/contracts/harness-driver.ts b/packages/paperclip-runner/src/contracts/harness-driver.ts
index 410496afce5..a520a146431 100644
--- a/packages/paperclip-runner/src/contracts/harness-driver.ts
+++ b/packages/paperclip-runner/src/contracts/harness-driver.ts
@@ -498,6 +498,7 @@ export interface HarnessSessionRecoveryResult {
}
export interface HarnessSession {
+ supportsTurnReasoning?(): boolean;
turnControlCapabilities?(): NativeTurnControlCapabilities | null;
ids(): {
driverSessionId: string;
@@ -511,6 +512,8 @@ export interface HarnessSession {
/** Set by orchestration only after successful provider-session recovery. */
continuation?: true;
requestedCollaborationMode?: "default" | "plan";
+ /** OpenCode/OpenRouter only. Applies to this turn, never subsequent turns. */
+ reasoningMode?: "default" | "disabled";
}): Promise<{
turnId: string;
effectiveCollaborationMode?: "default" | "plan";
diff --git a/packages/paperclip-runner/src/contracts/native-session-backend.ts b/packages/paperclip-runner/src/contracts/native-session-backend.ts
index f473b56a03e..251d10748e7 100644
--- a/packages/paperclip-runner/src/contracts/native-session-backend.ts
+++ b/packages/paperclip-runner/src/contracts/native-session-backend.ts
@@ -193,6 +193,8 @@ export interface NativeSession {
/** Set by orchestration only after successful provider-session recovery. */
continuation?: true;
requestedCollaborationMode?: "default" | "plan";
+ /** OpenCode/OpenRouter only. Applies to this turn, never subsequent turns. */
+ reasoningMode?: "default" | "disabled";
}): Promise<{
turnId: string;
effectiveCollaborationMode?: "default" | "plan";
diff --git a/packages/paperclip-runner/src/contracts/types.ts b/packages/paperclip-runner/src/contracts/types.ts
index 81bc3d24723..8a5dd2a49bd 100644
--- a/packages/paperclip-runner/src/contracts/types.ts
+++ b/packages/paperclip-runner/src/contracts/types.ts
@@ -41,6 +41,8 @@ export interface NativeSessionCapabilities {
runtimeRequestHandoff?: boolean;
goals?: boolean;
threadLineage?: boolean;
+ /** Can select reasoning for one turn without changing subsequent turns. */
+ perTurnReasoning?: boolean;
collaborationModes?: Array<"default" | "plan">;
unsupported?: string[];
}
diff --git a/packages/paperclip-runner/src/drivers/codex/app-server-transport.ts b/packages/paperclip-runner/src/drivers/codex/app-server-transport.ts
index 3e08de1f22f..a65138a435b 100644
--- a/packages/paperclip-runner/src/drivers/codex/app-server-transport.ts
+++ b/packages/paperclip-runner/src/drivers/codex/app-server-transport.ts
@@ -35,6 +35,7 @@ export type CodexServerRequestHandler = (
) => Promise>;
export interface CodexAppServerTransport {
+ supportsTurnReasoning?(): boolean;
/** Runner-owned live capability projection; absent on native Codex transports. */
turnControlCapabilities?(): NativeTurnControlCapabilities | null;
request(
diff --git a/packages/paperclip-runner/src/drivers/codex/codex-app-server-driver.context-control.test.ts b/packages/paperclip-runner/src/drivers/codex/codex-app-server-driver.context-control.test.ts
index cd9ee34129d..021c0ab99f7 100644
--- a/packages/paperclip-runner/src/drivers/codex/codex-app-server-driver.context-control.test.ts
+++ b/packages/paperclip-runner/src/drivers/codex/codex-app-server-driver.context-control.test.ts
@@ -45,6 +45,19 @@ import {
import { rehydrateRunnerdItemNotification } from "../../live/runnerd-codex-transport.js";
describe("Codex app-server Codex driver", () => {
+ it.each(["default", "disabled"] as const)("rejects unsupported reasoning %s without sending a turn RPC", async reasoningMode => {
+ const transport = new FakeCodexTransport();
+ const session = await makeDriver([transport]).openSession({
+ runId: "run-reasoning", normalizedSessionId: "normalized-reasoning", workingDirectory: WORKSPACE,
+ });
+ const message = { role: "user" as const, text: "Hi" };
+ await expect(session.startTurn({ message, reasoningMode })).rejects.toThrow("Per-turn reasoning is not supported");
+ expect(transport.calls.filter(call => call.method === "turn/start")).toHaveLength(0);
+ await session.startTurn({ message });
+ expect(transport.calls.filter(call => call.method === "turn/start")).toHaveLength(1);
+ await session.close({ reason: "fixture complete" });
+ });
+
it("retains one correlation-bound steering acknowledgement after the rehydrated runnerd echo", async () => {
const transport = new FakeCodexTransport("thread-root");
const session = await makeDriver([transport]).openSession({
diff --git a/packages/paperclip-runner/src/drivers/codex/codex-harness-session.ts b/packages/paperclip-runner/src/drivers/codex/codex-harness-session.ts
index 4efc9eec23d..983d9322eee 100644
--- a/packages/paperclip-runner/src/drivers/codex/codex-harness-session.ts
+++ b/packages/paperclip-runner/src/drivers/codex/codex-harness-session.ts
@@ -82,6 +82,10 @@ export class CodexHarnessSession
}
}
+ supportsTurnReasoning(): boolean {
+ return this.transport.supportsTurnReasoning?.() === true;
+ }
+
turnControlCapabilities() {
if (this.driverKind === "acpx_runtime") {
return this.transport.turnControlCapabilities?.() ?? { steering: false, queuedFollowUp: false };
@@ -157,11 +161,16 @@ export class CodexHarnessSession
/** Set by orchestration only after successful provider-session recovery. */
continuation?: true;
requestedCollaborationMode?: "default" | "plan";
+ /** OpenCode/OpenRouter only. Applies to this turn, never subsequent turns. */
+ reasoningMode?: "default" | "disabled";
}): Promise<{
turnId: string;
effectiveCollaborationMode: "default" | "plan";
}> {
this.assertProtocolIntegrity();
+ if (input.reasoningMode !== undefined && !this.supportsTurnReasoning()) {
+ throw new Error("Per-turn reasoning is not supported by this provider");
+ }
if (this.protocolFailed && this.protocolFailureCode) {
throw new NativeProviderTerminalFailure(this.protocolFailureCode, false, this.protocolFailureMessage ?? undefined);
}
@@ -231,6 +240,7 @@ export class CodexHarnessSession
try {
response = await this.transport.request("turn/start", {
threadId: this.opened.threadId,
+ ...(input.reasoningMode === undefined ? {} : { reasoningMode: input.reasoningMode }),
...(this.reasoningEffort ? { effort: this.reasoningEffort } : {}),
cwd: this.opened.context.workingDirectory,
permissions:
diff --git a/packages/paperclip-runner/src/drivers/opencode/opencode-server-driver.test.ts b/packages/paperclip-runner/src/drivers/opencode/opencode-server-driver.test.ts
index 59d740d9940..75032bd82c7 100644
--- a/packages/paperclip-runner/src/drivers/opencode/opencode-server-driver.test.ts
+++ b/packages/paperclip-runner/src/drivers/opencode/opencode-server-driver.test.ts
@@ -906,6 +906,8 @@ describe("OpenCodeServerDriver", () => {
expect(environment.keys).not.toContain("UNRELATED_SECRET");
expect(environment.keys).not.toContain("PAPERCLIP_PROVIDER_TRACE_PATH");
expect(environment.projectConfigDisabled).toBe("true");
+ expect(environment.modelsFetchDisabled).toBe("true");
+ expect(environment.defaultPluginsDisabled).toBe("true");
expect(mcpEvidence.tools).toEqual(
expect.arrayContaining(["paperclip_finish", "paperclip_block"]),
);
@@ -920,6 +922,46 @@ describe("OpenCodeServerDriver", () => {
expect(diagnostics.join("\n")).toContain("[REDACTED]");
});
+ it.each(["role-first", "role-last", "role-last-long"])("preserves incremental text and completion with identity checks (%s)", async (roleOrder) => {
+ await chmod(fixture, 0o755);
+ const root = await mkdtemp(join(tmpdir(), "paperclip-opencode-stream-"));
+ const workspace = await mkdtemp(join(tmpdir(), "paperclip-opencode-stream-workspace-"));
+ roots.push(root, workspace);
+ const driver = new OpenCodeServerDriver({
+ model: "openrouter/deepseek/deepseek-v4-flash-0731",
+ runtimeDirectory: root,
+ command: fixture,
+ environment: { PATH: process.env.PATH },
+ });
+ const session = await driver.openSession({ runId: "stream", normalizedSessionId: "stream", workingDirectory: workspace });
+ try {
+ await session.startTurn({ message: { role: "user", text: `incremental-stream ${roleOrder}` } });
+ const deltas: string[] = [];
+ const reasoning: string[] = [];
+ const events: PrpEvent[] = [];
+ for await (const event of session.events()) {
+ events.push(event);
+ if (event.eventType === "item.delta" && event.payload.kind === "reasoning") reasoning.push(String(event.payload.text));
+ if (event.eventType === "item.delta" && event.payload.kind === "agentMessage") {
+ deltas.push(String(event.payload.text));
+ // The fixture cannot finish its turn until a chunk reaches us.
+ if (deltas.length === 1) {
+ expect(await readFile(join(root, "stream", "data", "stream-released"), "utf8").catch(() => null)).toBeNull();
+ }
+ await writeFile(join(root, "stream", "data", "release-stream"), "ready");
+ }
+ if (TURN_TERMINAL_EVENT_TYPES.has(event.eventType)) break;
+ }
+ const expectedText = `Hi ${"👋".repeat(roleOrder === "role-last-long" ? 120 : 2)}`;
+ expect(deltas).toEqual(roleOrder === "role-first" ? ["Hi ", "👋", "👋"] : [expectedText]);
+ expect(reasoning).toEqual(["Thinking"]);
+ expect(events.filter(event => event.eventType === "item.completed" && event.payload.kind === "agentMessage")
+ .map(event => event.payload.text)).toEqual([expectedText]);
+ } finally {
+ await session.close({ reason: "stream test complete" });
+ }
+ });
+
it("maps OpenCode's normal abort error to cancellation without a false provider failure notice", async () => {
await chmod(fixture, 0o755);
const root = await mkdtemp(join(tmpdir(), "paperclip-opencode-abort-"));
@@ -1497,7 +1539,7 @@ describe("OpenCodeServerDriver", () => {
await session.close({ reason: "test" });
});
- it("sends only the authoritative wake envelope when resuming an existing provider session", async () => {
+ it("resends system instructions without repeating the task envelope on recovery", async () => {
await chmod(fixture, 0o755);
const root = await mkdtemp(
join(tmpdir(), "paperclip-opencode-resume-context-"),
@@ -1553,7 +1595,7 @@ describe("OpenCodeServerDriver", () => {
],
});
expect(submittedPrompt).not.toHaveProperty("tools");
- expect(submittedPrompt).not.toHaveProperty("system");
+ expect(submittedPrompt).toHaveProperty("system", "large original system context");
await recovered!.session!.close({ reason: "recovery-test" });
});
@@ -1942,9 +1984,11 @@ describe("OpenCodeServerDriver", () => {
"*": permissionMode,
external_directory: { "*": "deny", [`${workspace}/**`]: "allow" },
});
+ expect(config.autoupdate).toBe(false);
expect(config.provider.openrouter.models).toHaveProperty(
"deepseek/deepseek-v4-flash-0731",
);
+ expect(config.provider.openrouter.models["deepseek/deepseek-v4-flash-0731"].options).toBeUndefined();
expect(config.permission.paperclip_finish).toBeUndefined();
expect(config.permission["paperclip_*"]).toBe("allow");
await session.close({ reason: "permission mode test complete" });
@@ -2440,6 +2484,9 @@ describe("OpenCodeServerDriver", () => {
afterSpawn: () => {
commandLifecycle.push("after");
},
+ afterExit: () => {
+ commandLifecycle.push("exit");
+ },
},
fetch: async (input, init) => {
if (
@@ -2465,9 +2512,31 @@ describe("OpenCodeServerDriver", () => {
expect(spawns.length).toBeGreaterThanOrEqual(2);
expect(spawns.length).toBeLessThanOrEqual(3);
expect(commandLifecycle).toEqual(
- Array.from({ length: spawns.length }, () => ["before", "after"]).flat(),
+ [...Array.from({ length: spawns.length - 1 }, () => ["before", "after", "exit"]).flat(), "before", "after"],
);
await session.close({ reason: "test" });
+ expect(commandLifecycle).toEqual(
+ Array.from({ length: spawns.length }, () => ["before", "after", "exit"]).flat(),
+ );
+ });
+
+ it("releases the executable binding when spawning fails before health checks", async () => {
+ const root = await mkdtemp(join(tmpdir(), "paperclip-opencode-spawn-error-"));
+ roots.push(root);
+ const lifecycle: string[] = [];
+ const driver = new OpenCodeServerDriver({
+ model: "openrouter/deepseek/deepseek-v4-flash-0731",
+ runtimeDirectory: root,
+ command: join(root, "missing-executable"),
+ commandLifecycle: {
+ beforeSpawn: () => { lifecycle.push("before"); },
+ afterSpawn: () => { lifecycle.push("spawned"); },
+ afterExit: () => { lifecycle.push("released"); },
+ },
+ });
+ await expect(driver.openSession({ runId: "spawn-error", normalizedSessionId: "spawn-error",
+ workingDirectory: root })).rejects.toThrow("ENOENT");
+ expect(lifecycle).toEqual(["before", "released"]);
});
it("reports startup exit details with stderr captured after spawn and redacted", async () => {
diff --git a/packages/paperclip-runner/src/drivers/opencode/opencode-server-driver.ts b/packages/paperclip-runner/src/drivers/opencode/opencode-server-driver.ts
index 81abbfd3f06..4ce6a5b9dd9 100644
--- a/packages/paperclip-runner/src/drivers/opencode/opencode-server-driver.ts
+++ b/packages/paperclip-runner/src/drivers/opencode/opencode-server-driver.ts
@@ -19,6 +19,8 @@ import { getCACertificates } from "node:tls";
import { pipeline } from "node:stream/promises";
import { dirname, isAbsolute, join, relative, resolve, sep } from "node:path";
+import { parseOpenCodeReasoningMode, type OpenCodeReasoningMode } from "./reasoning-mode.js";
+
import {
CODEX_SKILLLESS_BASE_INSTRUCTIONS,
createCodexTaskEnvelope,
@@ -176,6 +178,7 @@ export interface OpenCodeServerDriverOptions {
commandLifecycle?: {
beforeSpawn(): void;
afterSpawn(): void;
+ afterExit?(): void;
};
runtimeDirectory: string;
systemInstructions?: string;
@@ -479,6 +482,7 @@ class OpenCodeHarnessSession implements HarnessSession {
Array>
>();
readonly #partText = new Map();
+ readonly #streamingParts = new Map>();
readonly #completedTextPartIds = new Set();
readonly #completedReasoningPartIds = new Set();
readonly #completedTextParts: Array<{
@@ -640,12 +644,22 @@ class OpenCodeHarnessSession implements HarnessSession {
.finally(() => this.#pumpEvents());
}
+ supportsTurnReasoning(): boolean {
+ return this.#model.startsWith("openrouter/");
+ }
+
async startTurn(input: {
message: NativeUserMessage;
+ reasoningMode?: OpenCodeReasoningMode;
}): Promise<{ turnId: string }> {
if (this.#activeTurnId !== null)
throw new Error("OpenCode session already has an active turn");
+ const reasoningMode = parseOpenCodeReasoningMode(input.reasoningMode);
+ if (input.reasoningMode !== undefined && !this.#model.startsWith("openrouter/")) {
+ throw new Error("Per-turn OpenCode reasoning is supported only for OpenRouter models");
+ }
const turnId = `turn-${randomBytes(12).toString("hex")}`;
+ this.#streamingParts.clear();
this.#activeTurnId = turnId;
this.#emit("turn.submitted", {
envelopeSchema: this.#taskEnvelope.schema,
@@ -655,12 +669,9 @@ class OpenCodeHarnessSession implements HarnessSession {
this.#emit("turn.started", { status: "inProgress" }, { turnId });
const [providerID, ...modelParts] = this.#model.split("/");
const modelID = modelParts.join("/");
- // A resumed OpenCode provider session already retains the original system
- // instructions and task envelope in its conversation. Repeating both on
- // every Paperclip continuation can overflow smaller context windows and
- // OpenCode then completes with `finish: unknown` and zero tokens. The
- // native model envelope still carries the authoritative wake delta,
- // interaction responses, completion contract, and current issue context.
+ // Keep the task envelope only on the initial task-mode wake. OpenCode
+ // rebuilds system instructions from the latest user message, so system
+ // instructions must still accompany every prompt (including recovery).
const prompt = this.#sendFullContext
? JSON.stringify({
task: this.#taskEnvelope,
@@ -682,12 +693,15 @@ class OpenCodeHarnessSession implements HarnessSession {
// at this HTTP boundary.
providerID,
modelID,
+ // Always select the empty default variant on an ordinary turn so a
+ // previous disabled turn cannot change this turn's provider defaults.
+ ...(providerID === "openrouter"
+ ? { variant: reasoningMode === "disabled" ? "paperclip-no-reasoning" : "paperclip-default" }
+ : {}),
// Keep question enabled in the isolated config. OpenCode 1.18.32
// turns a prompt's deprecated `tools` map into replacement session
// permissions, so a sparse override here discards the session policy.
- ...(this.#sendFullContext
- ? { system: this.#systemInstructions }
- : {}),
+ system: this.#systemInstructions,
parts: [{ type: "text", text: prompt }],
}),
},
@@ -1470,9 +1484,11 @@ class OpenCodeHarnessSession implements HarnessSession {
const event = record(value);
const properties = record(event.properties);
const type = text(event.type);
- const eventId = text(event.id, canonicalJson(value));
- if (this.#seenProviderEvents.has(eventId)) return;
- this.#seenProviderEvents.add(eventId);
+ // Id-less deltas are additive: two identical chunks can be legitimate
+ // adjacent tokens. Content deduplication would silently lose the second.
+ const eventId = text(event.id) || (type === "message.part.delta" ? null : canonicalJson(value));
+ if (eventId !== null && this.#seenProviderEvents.has(eventId)) return;
+ if (eventId !== null) this.#seenProviderEvents.add(eventId);
if (this.#seenProviderEvents.size > 10_000)
this.#seenProviderEvents.delete(
this.#seenProviderEvents.values().next().value!,
@@ -1565,8 +1581,16 @@ class OpenCodeHarnessSession implements HarnessSession {
);
return;
}
- if (type === "message.part.updated") {
- const part = record(properties.part);
+ if (type === "message.part.updated" || type === "message.part.delta") {
+ let part = record(properties.part);
+ if (type === "message.part.delta") {
+ const previous = this.#streamingParts.get(text(properties.partID));
+ // A delta has no role or part type. Only stream text for an observed
+ // part with matching message identity; never infer assistant identity.
+ if (!previous || properties.field !== "text" || typeof properties.delta !== "string"
+ || text(previous.messageID, text(previous.messageId)) !== text(properties.messageID)) return;
+ part = { ...previous, text: text(previous.text) + properties.delta };
+ }
const messageId = text(part.messageID, text(part.messageId));
if (!messageId) return;
// Resolve the turn this message actually belongs to, not whichever
@@ -1574,11 +1598,20 @@ class OpenCodeHarnessSession implements HarnessSession {
// must stay attributed to the turn that created that message.
const owningTurnId = this.#messageTurnIds.get(messageId) ?? turnId;
if (!owningTurnId) return;
+ if (owningTurnId === turnId && text(part.id) && ["text", "reasoning"].includes(text(part.type))) {
+ this.#streamingParts.set(text(part.id), part);
+ }
const role = this.#messageRoles.get(messageId);
if (role === "assistant") this.#emitAssistantPart(part, owningTurnId);
else if (role === undefined) {
const pending = this.#pendingMessageParts.get(messageId) ?? [];
- if (pending.length < 100) pending.push(part);
+ // Text deltas are cumulative by this point. Keep the newest snapshot
+ // (including completion metadata), not the first 100 token chunks.
+ const previousIndex = ["text", "reasoning"].includes(text(part.type)) && text(part.id)
+ ? pending.findIndex(candidate => candidate.id === part.id && candidate.type === part.type)
+ : -1;
+ if (previousIndex >= 0) pending[previousIndex] = part;
+ else if (pending.length < 100) pending.push(part);
this.#pendingMessageParts.set(messageId, pending);
}
return;
@@ -2233,12 +2266,23 @@ async function startRuntime(input: {
input.trace?.addSensitiveValues([providerProxy.token]);
}
let child: ChildProcess | undefined;
+ let launchPrepared = false;
+ const releaseExecutable = () => {
+ if (!launchPrepared) return;
+ launchPrepared = false;
+ try {
+ input.options.commandLifecycle?.afterExit?.();
+ } catch (error) {
+ input.options.onDiagnostic?.(redact(`OpenCode executable cleanup failed: ${String(error)}`, sensitiveValues));
+ }
+ };
try {
const config = {
$schema: "https://opencode.ai/config.json",
model: input.options.model,
small_model: input.options.model,
share: "disabled",
+ autoupdate: false,
// The configured entry is already composed exactly once into the session
// system prompt; siblings remain available through the read-only root.
instructions: [],
@@ -2257,7 +2301,15 @@ async function startRuntime(input: {
},
} : {}),
models: {
- [providerModelId]: { name: providerModelId },
+ [providerModelId]: {
+ name: providerModelId,
+ ...(modelProvider === "openrouter"
+ ? { variants: {
+ "paperclip-default": {},
+ "paperclip-no-reasoning": { reasoning: { enabled: false } },
+ } }
+ : {}),
+ },
},
},
},
@@ -2307,6 +2359,11 @@ async function startRuntime(input: {
XDG_DATA_HOME: dataHome,
XDG_CACHE_HOME: cacheHome,
OPENCODE_DISABLE_PROJECT_CONFIG: "true",
+ // The runner pins the executable, model and assigned MCP tools. A
+ // fresh isolated cache must not fetch a catalog or install unrelated
+ // default plugins before it can submit the first prompt.
+ OPENCODE_DISABLE_MODELS_FETCH: "true",
+ OPENCODE_DISABLE_DEFAULT_PLUGINS: "true",
OPENCODE_SERVER_USERNAME: username,
OPENCODE_SERVER_PASSWORD: password,
...(providerProxy ? { NO_PROXY: [input.options.environment?.no_proxy ?? input.options.environment?.NO_PROXY, "127.0.0.1", "localhost"].filter(Boolean).join(",") } : {}),
@@ -2319,6 +2376,7 @@ async function startRuntime(input: {
stdio[input.options.commandFd] = input.options.commandFd;
}
input.options.commandLifecycle?.beforeSpawn();
+ launchPrepared = true;
child = spawn(
input.options.command ?? resolvePinnedOpenCodeCommand(),
["serve", "--hostname", "127.0.0.1", "--port", String(port)],
@@ -2329,6 +2387,7 @@ async function startRuntime(input: {
detached: globalThis.process.platform !== "win32" && isolateProcessGroup,
},
);
+ child.once("exit", releaseExecutable);
if (child.pid !== undefined) {
try {
input.options.commandLifecycle?.afterSpawn();
@@ -2425,6 +2484,7 @@ async function startRuntime(input: {
} catch {
providerChild.kill("SIGKILL");
}
+ await waitForExit(providerChild, 2_000);
}
await rm(join(configHome, "opencode", "opencode.json"), {
force: true,
@@ -2441,6 +2501,8 @@ async function startRuntime(input: {
await providerProxy?.close();
await bridge.close().catch(() => {});
child?.kill("SIGKILL");
+ if (child?.pid) await waitForExit(child, 2_000);
+ if (!child?.pid || child.exitCode !== null || child.signalCode !== null) releaseExecutable();
await rm(join(configHome, "opencode", "opencode.json"), {
force: true,
}).catch(() => undefined);
diff --git a/packages/paperclip-runner/src/drivers/opencode/prepared-context-transport.test.ts b/packages/paperclip-runner/src/drivers/opencode/prepared-context-transport.test.ts
index dd6f34e8677..c4710a971e9 100644
--- a/packages/paperclip-runner/src/drivers/opencode/prepared-context-transport.test.ts
+++ b/packages/paperclip-runner/src/drivers/opencode/prepared-context-transport.test.ts
@@ -30,6 +30,7 @@ describe("prepared OpenCode context transport", () => {
runtimeDirectory: root,
command: fixture,
conversationMode: "prepared",
+ systemInstructions: "Prepared agent instructions.",
environment: { PATH: process.env.PATH, OPENROUTER_API_KEY: "fixture-key" },
fetch: async (input, init) => {
if (String(input).endsWith("/prompt_async"))
@@ -45,6 +46,7 @@ describe("prepared OpenCode context transport", () => {
});
await session.startTurn({
message: { role: "user", text: "prepared OpenCode initial wake" },
+ reasoningMode: "disabled",
});
for await (const event of session.events()) {
if (event.eventType === "turn.completed") break;
@@ -67,6 +69,8 @@ describe("prepared OpenCode context transport", () => {
[{ type: "text", text: "completion-only OpenCode continuation" }],
]);
expect(JSON.stringify(submitted)).not.toContain('"task"');
+ expect(submitted.map(body => body.variant)).toEqual(["paperclip-no-reasoning", "paperclip-default"]);
+ expect(submitted.map(body => body.system)).toEqual(["Prepared agent instructions.", "Prepared agent instructions."]);
await recovered!.session!.close({ reason: "prepared-test-recovery" });
});
});
diff --git a/packages/paperclip-runner/src/drivers/opencode/reasoning-mode.test.ts b/packages/paperclip-runner/src/drivers/opencode/reasoning-mode.test.ts
new file mode 100644
index 00000000000..db5e747d46a
--- /dev/null
+++ b/packages/paperclip-runner/src/drivers/opencode/reasoning-mode.test.ts
@@ -0,0 +1,63 @@
+import { chmod, mkdtemp, readFile, rm } from "node:fs/promises";
+import { tmpdir } from "node:os";
+import { join, resolve } from "node:path";
+import { describe, expect, it } from "vitest";
+import { OpenCodeServerDriver } from "./opencode-server-driver.js";
+import { parseOpenCodeReasoningMode } from "./reasoning-mode.js";
+import { openCodeProxyTaskEnvelope } from "../../cli/opencode-proxy-task-envelope.js";
+
+describe("per-turn OpenCode reasoning", () => {
+ it.each([undefined, "default"])("preserves provider defaults for %s", (value) => {
+ expect(parseOpenCodeReasoningMode(value)).toBe("default");
+ });
+ it.each(["", "false", null, false, 0])("rejects malformed mode %s", (value) => {
+ expect(() => parseOpenCodeReasoningMode(value)).toThrow("turn.reasoningMode");
+ });
+ it("selects reasoning independently on successive turns and sends instructions once", async () => {
+ const root = await mkdtemp(join(tmpdir(), "paperclip-turn-reasoning-"));
+ const fixture = resolve("test/fixtures/fake-opencode-server.mjs");
+ await chmod(fixture, 0o755);
+ const submitted: Array<{ variant?: string; system?: string; parts: Array<{ text: string }> }> = [];
+ const instructions = "Unique shared agent instructions. Keep the requested scope.";
+ const driver = new OpenCodeServerDriver({
+ model: "openrouter/deepseek/deepseek-v4-flash-0731",
+ runtimeDirectory: root,
+ command: fixture,
+ systemInstructions: instructions,
+ taskEnvelope: openCodeProxyTaskEnvelope({ baseInstructions: instructions,
+ completionContract: { revision: "17", criterionIds: ["objective"] } }),
+ environment: { PATH: process.env.PATH, OPENROUTER_API_KEY: "fixture-key" },
+ fetch: async (input, init) => {
+ if (String(input).endsWith("/prompt_async")) submitted.push(JSON.parse(String(init?.body)));
+ return fetch(input, init);
+ },
+ });
+ let session: Awaited> | undefined;
+ try {
+ session = await driver.openSession({ runId: "turn-reasoning", normalizedSessionId: "reasoning", workingDirectory: root });
+ await expect(session.startTurn({ message: { role: "user", text: "invalid" }, reasoningMode: "bad" as never })).rejects.toThrow("turn.reasoningMode");
+ expect(submitted).toHaveLength(0);
+ for (const reasoningMode of [undefined, "disabled", undefined] as const) {
+ await session.startTurn({ message: { role: "user", text: "Reply Hi." }, reasoningMode });
+ for await (const event of session.events()) if (event.eventType === "turn.completed") break;
+ }
+ expect(submitted.map(body => body.variant)).toEqual(["paperclip-default", "paperclip-no-reasoning", "paperclip-default"]);
+ expect(submitted[0]!.system).toBe(instructions);
+ for (const body of submitted) {
+ expect(body.system).toBe(instructions);
+ expect(JSON.stringify(body).split(instructions)).toHaveLength(2);
+ }
+ const envelope = JSON.parse(submitted[0]!.parts[0].text);
+ expect(envelope.task.completionContract.revision).toBe("17");
+ expect(envelope.task.constraints).toEqual(["Use Paperclip MCP tools for semantic operations."]);
+ const config = JSON.parse(await readFile(join(root, "reasoning/config/opencode/opencode.json"), "utf8"));
+ expect(config.provider.openrouter.models["deepseek/deepseek-v4-flash-0731"]).toMatchObject({
+ variants: { "paperclip-default": {}, "paperclip-no-reasoning": { reasoning: { enabled: false } } },
+ });
+ expect(config.provider.openrouter.models["deepseek/deepseek-v4-flash-0731"].options).toBeUndefined();
+ } finally {
+ await session?.close();
+ await rm(root, { recursive: true, force: true });
+ }
+ });
+});
diff --git a/packages/paperclip-runner/src/drivers/opencode/reasoning-mode.ts b/packages/paperclip-runner/src/drivers/opencode/reasoning-mode.ts
new file mode 100644
index 00000000000..003589ead39
--- /dev/null
+++ b/packages/paperclip-runner/src/drivers/opencode/reasoning-mode.ts
@@ -0,0 +1,8 @@
+export type OpenCodeReasoningMode = "default" | "disabled";
+
+/** Opt-in only. Omitting the setting preserves the provider's own defaults. */
+export function parseOpenCodeReasoningMode(value: unknown): OpenCodeReasoningMode {
+ if (value === undefined || value === "default") return "default";
+ if (value === "disabled") return "disabled";
+ throw new Error("turn.reasoningMode must be default or disabled");
+}
diff --git a/packages/paperclip-runner/src/live/live-session.ts b/packages/paperclip-runner/src/live/live-session.ts
index b79755a6685..c36b8036346 100644
--- a/packages/paperclip-runner/src/live/live-session.ts
+++ b/packages/paperclip-runner/src/live/live-session.ts
@@ -1514,7 +1514,7 @@ export class CapabilityLiveSession {
async sendMessage(
message: string,
/** Launch-only diagnostics may opt out; qualification campaigns must not. */
- options: { allowMissingUsage?: boolean } = {},
+ options: { allowMissingUsage?: boolean; reasoningMode?: "default" | "disabled" } = {},
): Promise {
const value = message.trim();
if (value.length === 0) throw new Error("Capability live messages cannot be empty");
@@ -1617,6 +1617,7 @@ export class CapabilityLiveSession {
permissions: CODEX_PERMISSION_PROFILE,
runtimeWorkspaceRoots: [this.#config.workingDirectory],
input: [userInput(value)],
+ ...(options.reasoningMode === undefined ? {} : { reasoningMode: options.reasoningMode }),
});
} catch (error) {
if (this.#pendingTurnAdmission !== admission) {
diff --git a/packages/paperclip-runner/src/live/runnerd-codex-transport.test.ts b/packages/paperclip-runner/src/live/runnerd-codex-transport.test.ts
index a024eb20d92..debd0c6e3fe 100644
--- a/packages/paperclip-runner/src/live/runnerd-codex-transport.test.ts
+++ b/packages/paperclip-runner/src/live/runnerd-codex-transport.test.ts
@@ -7951,7 +7951,10 @@ it.each([true, false])("preserves prepared OpenCode cleanup errors (primary fail
expect((failure as Error).message).not.toContain("fixture-secret");
});
-it("preserves prepared input and completion feedback through runnerd and the real OpenCode proxy boundary", async () => {
+it.each([
+ { model: "openrouter/deepseek/deepseek-v4-flash-0731", supportsReasoning: true },
+ { model: "anthropic/claude-sonnet-4", supportsReasoning: false },
+])("preserves prepared input and completion feedback through runnerd and the real OpenCode proxy boundary ($model)", async ({ model, supportsReasoning }) => {
const root = await mkdtemp(join(tmpdir(), "runnerd-prepared-opencode-"));
// GitHub-hosted Linux toolcache Node can be group-writable, unlike the AWS
// fleet. Qualify an owned copy with strict permissions, never chmod the host
@@ -8006,7 +8009,7 @@ it("preserves prepared input and completion feedback through runnerd and the rea
const driver = new CodexAppServerDriver({
taskEnvelope: task,
conversationMode: "prepared",
- model: "openrouter/deepseek/deepseek-v4-flash-0731",
+ model,
transportFactory: () => bundle.transport,
workingDirectoryAuthority: "remote_runner",
environment: { PAPERCLIP_WORKSPACE_CWD: root },
@@ -8025,7 +8028,15 @@ it("preserves prepared input and completion feedback through runnerd and the rea
await withPreparedOpenCodeCleanup({
run: async () => {
session = await driver.openSession({ runId: "prepared-opencode", normalizedSessionId: "prepared-opencode", workingDirectory: root });
- await session.startTurn({ message: { role: "user", text: prepared } });
+ expect(bundle.transport.supportsTurnReasoning?.()).toBe(supportsReasoning);
+ const firstTurn = { message: { role: "user" as const, text: prepared } };
+ if (!supportsReasoning) {
+ await expect(session.startTurn({ ...firstTurn, reasoningMode: "disabled" }))
+ .rejects.toThrow("Per-turn reasoning is not supported");
+ await expect(bundle.transport.request("turn/start", { input: [{ type: "text", text: prepared }], reasoningMode: "disabled" }))
+ .rejects.toThrow("requires an OpenCode/OpenRouter model");
+ }
+ await session.startTurn({ ...firstTurn, ...(supportsReasoning ? { reasoningMode: "disabled" as const } : {}) });
const events: PrpEvent[] = [];
for await (const event of session.events()) {
events.push(event);
@@ -8037,10 +8048,16 @@ it("preserves prepared input and completion feedback through runnerd and the rea
expect(sessionRoots).toHaveLength(1);
const requests = (await readFile(join(runtime, sessionRoots[0]!.name, "data/fake-prompt-requests.ndjson"), "utf8")).trim().split("\n").map((line) => JSON.parse(line));
expect(requests.map((request) => request.parts)).toEqual([[{ type: "text", text: prepared }]]);
+ expect(requests[0].variant).toBe(supportsReasoning ? "paperclip-no-reasoning" : undefined);
const outcomes = JSON.parse(await readFile(join(runtime, sessionRoots[0]!.name, "data/fake-completion-feedback.json"), "utf8"));
expect(outcomes).toHaveLength(2);
expect(outcomes[0].result).toMatchObject({ isError: true, content: [{ text: expect.stringContaining("required document link") }] });
expect(outcomes[1].result).toMatchObject({ content: [{ text: expect.stringContaining(feedback) }] });
+ await session.startTurn({ message: { role: "user", text: "Reply Hi." } });
+ for await (const event of session.events()) if (event.eventType === "turn.completed") break;
+ const allRequests = (await readFile(join(runtime, sessionRoots[0]!.name, "data/fake-prompt-requests.ndjson"), "utf8")).trim().split("\n").map((line) => JSON.parse(line));
+ expect(allRequests.map(request => request.variant)).toEqual(supportsReasoning
+ ? ["paperclip-no-reasoning", "paperclip-default"] : [undefined, undefined]);
},
closeSession: async () => { await session?.close(); },
closeTransport: () => bundle.transport.close(),
diff --git a/packages/paperclip-runner/src/live/runnerd-codex-transport.ts b/packages/paperclip-runner/src/live/runnerd-codex-transport.ts
index 08a8886f3b9..71103812f2a 100644
--- a/packages/paperclip-runner/src/live/runnerd-codex-transport.ts
+++ b/packages/paperclip-runner/src/live/runnerd-codex-transport.ts
@@ -3538,6 +3538,20 @@ class DurablePrpCodexTransport implements CodexAppServerTransport {
#providerIdentity: Record | null = null;
#turnControls: NativeTurnControlCapabilities = { steering: false, queuedFollowUp: false };
+ supportsTurnReasoning(): boolean {
+ if (this.options.provider !== "opencode") return false;
+ // Use the durable selected model on both fresh and recovered sessions.
+ // Older states may retain the provider only in the preparation command.
+ const state = this.#core?.store.state;
+ const template = this.#runAttachTemplate ?? state?.runAttachTemplate
+ ?? state?.commands.findLast(command =>
+ (command.type === "run.prepare" || command.type === "run.attach")
+ && record(command.payload).provider !== undefined,
+ )?.payload;
+ const model = record(record(template).provider).model;
+ return typeof model === "string" && model.startsWith("openrouter/");
+ }
+
turnControlCapabilities(): NativeTurnControlCapabilities | null {
if (this.options.provider !== "acpx") return null;
if (this.#closed || this.#failure) return { steering: false, queuedFollowUp: false };
@@ -5883,6 +5897,11 @@ class DurablePrpCodexTransport implements CodexAppServerTransport {
.filter((item) => item.type !== "skill")
.map((item) => (typeof item.text === "string" ? item.text : ""))
.join("\n");
+ const reasoningMode = params.reasoningMode;
+ if (reasoningMode !== undefined && (
+ !this.supportsTurnReasoning()
+ || (reasoningMode !== "default" && reasoningMode !== "disabled")
+ )) throw new Error("turn.reasoningMode requires an OpenCode/OpenRouter model and must be default or disabled");
const skills = resolveRunnerdCodexSkillInputs(
input.filter((item) => item.type === "skill"),
this.options.runtimeContext ?? null,
@@ -5916,6 +5935,7 @@ class DurablePrpCodexTransport implements CodexAppServerTransport {
"turn.start",
{
text: message,
+ ...(reasoningMode === undefined ? {} : { reasoningMode }),
...(skills.length ? { skills } : {}),
turnId: pendingTurnId,
},
diff --git a/packages/paperclip-runner/test/fixtures/fake-opencode-server.mjs b/packages/paperclip-runner/test/fixtures/fake-opencode-server.mjs
index 69bd35a8e4e..692fd9c2ebd 100644
--- a/packages/paperclip-runner/test/fixtures/fake-opencode-server.mjs
+++ b/packages/paperclip-runner/test/fixtures/fake-opencode-server.mjs
@@ -99,6 +99,8 @@ await writeFile(
home: process.env.HOME,
configHome: process.env.XDG_CONFIG_HOME,
projectConfigDisabled: process.env.OPENCODE_DISABLE_PROJECT_CONFIG,
+ modelsFetchDisabled: process.env.OPENCODE_DISABLE_MODELS_FETCH,
+ defaultPluginsDisabled: process.env.OPENCODE_DISABLE_DEFAULT_PLUGINS,
}),
);
@@ -400,8 +402,7 @@ const server = createServer(async (request, response) => {
const promptPayload = JSON.parse(Buffer.concat(chunks).toString("utf8"));
await appendFile(join(process.env.XDG_DATA_HOME, "fake-prompt-requests.ndjson"), JSON.stringify(promptPayload) + "\n");
if (
- promptPayload.providerID !== "openrouter" ||
- promptPayload.modelID !== "deepseek/deepseek-v4-flash-0731" ||
+ `${promptPayload.providerID}/${promptPayload.modelID}` !== runtimeConfig.model ||
"model" in promptPayload
) {
return json(response, 400, {
@@ -427,11 +428,56 @@ const server = createServer(async (request, response) => {
},
},
});
+ emit({
+ type: "message.part.delta",
+ properties: {
+ sessionID: session.id,
+ messageID: "message-late-source",
+ partID: "part-late-straggler",
+ field: "text",
+ delta: "late straggler text must not reach the next turn",
+ },
+ });
}
}
setTimeout(async () => {
await callFirstPaperclipTool();
const parsedPrompt = parsedPromptText(promptPayload);
+ if (String(parsedPrompt.message ?? "").includes("incremental-stream")) {
+ const messageID = `stream-message-${promptTurnSeq}`;
+ const part = { id: `stream-part-${promptTurnSeq}`, messageID, sessionID: session.id, type: "text", text: "" };
+ const reasoning = { ...part, id: `${part.id}-reasoning`, type: "reasoning" };
+ const update = (value) => emit({ type: "message.part.updated", properties: { sessionID: session.id, part: value } });
+ const delta = (value, text, extra = {}) => emit({ type: "message.part.delta", properties: { sessionID: session.id, messageID, partID: value.id, field: "text", delta: text, ...extra } });
+ const announceRole = () => emit({ type: "message.updated", properties: { sessionID: session.id, info: { id: messageID, role: "assistant" } } });
+ const roleFirst = String(parsedPrompt.message).includes("role-first");
+ const longResponse = String(parsedPrompt.message).includes("role-last-long");
+ if (roleFirst) announceRole();
+ // Parts and deltas can precede the message's role announcement.
+ update(part);
+ update(reasoning);
+ delta(reasoning, "Thinking");
+ delta(part, "Hi ");
+ const emojiCount = longResponse ? 120 : 2;
+ for (let i = 0; i < emojiCount; i++) delta(part, "👋");
+ delta(part, "wrong message", { messageID: "unrelated-message" });
+ delta(part, "wrong session", { sessionID: "unrelated-session" });
+ delta(part, "wrong field", { field: "output" });
+ delta(part, "unknown part", { partID: "unknown-part" });
+ const finalPart = { ...part, text: `Hi ${"👋".repeat(emojiCount)}`, time: { start: 1, end: 2 } };
+ if (longResponse) update(finalPart);
+ if (!roleFirst) announceRole();
+ // The consumer must observe a delta before the completed snapshot
+ // exists. This handshake makes the streaming regression deterministic.
+ for (let i = 0; i < 500; i++) {
+ if (await readFile(join(process.env.XDG_DATA_HOME, "release-stream"), "utf8").catch(() => "") === "ready") break;
+ await new Promise(resolve => setTimeout(resolve, 10));
+ }
+ await writeFile(join(process.env.XDG_DATA_HOME, "stream-released"), "ready");
+ update(finalPart);
+ emit({ type: "session.idle", properties: { sessionID: session.id } });
+ return;
+ }
if (String(parsedPrompt.message ?? "").includes("native-question")) {
pendingQuestion = nativeQuestion();
emit({
diff --git a/server/src/__tests__/heartbeat-responsible-user-invariant.test.ts b/server/src/__tests__/heartbeat-responsible-user-invariant.test.ts
index 59c3cb2c419..54bcebe2005 100644
--- a/server/src/__tests__/heartbeat-responsible-user-invariant.test.ts
+++ b/server/src/__tests__/heartbeat-responsible-user-invariant.test.ts
@@ -23,6 +23,7 @@ import {
import { heartbeatService } from "../services/heartbeat.ts";
import { runningProcesses } from "../adapters/index.ts";
import { drainHeartbeatRunsToQuiescence } from "./helpers/drain-heartbeat-runs.js";
+import { createLifecycleDriver } from "../services/agent-lifecycle-driver.js";
const mockAdapterExecute = vi.hoisted(() =>
vi.fn(async () => ({
@@ -159,6 +160,27 @@ describeEmbeddedPostgres("heartbeat responsible-user invariant", () => {
return { companyId, ownerUserId, agentId };
}
+ it("dispatches saved work on readiness with company scoping and normal admission checks", async () => {
+ const { companyId, agentId, ownerUserId } = await seedCompany();
+ await db.update(agents).set({ status: "paused", lifecycleState: "verifying" }).where(eq(agents.id, agentId));
+ const [run] = await db.insert(heartbeatRuns).values({ companyId, agentId, status: "queued",
+ invocationSource: "on_demand", contextSnapshot: { responsibleUserId: ownerUserId } }).returning();
+ const driver = createLifecycleDriver(db, {} as never);
+ const currentAgent = async () => (await db.select().from(agents).where(eq(agents.id, agentId)))[0];
+ await driver.onReady!(await currentAgent());
+ expect(mockAdapterExecute).not.toHaveBeenCalled();
+ expect((await db.select().from(heartbeatRuns).where(eq(heartbeatRuns.id, run.id)))[0].status).toBe("queued");
+
+ await db.update(agents).set({ status: "idle", lifecycleState: "ready" }).where(eq(agents.id, agentId));
+ await heartbeat.resumeQueuedRunsForAgent(randomUUID(), agentId);
+ expect(mockAdapterExecute).not.toHaveBeenCalled();
+ expect((await db.select().from(heartbeatRuns).where(eq(heartbeatRuns.id, run.id)))[0].status).toBe("queued");
+
+ await driver.onReady!(await currentAgent());
+ expect(await waitForRun(db, run.id)).toMatchObject({ status: "succeeded", responsibleUserId: ownerUserId });
+ expect(mockAdapterExecute).toHaveBeenCalledOnce();
+ });
+
it("dispatches an interrupted queue under the clicking operator through the real startup path", async () => {
const { companyId, agentId, ownerUserId } = await seedCompany();
const operatorId = `operator-${randomUUID()}`, issueId = randomUUID(), commentId = randomUUID(), queueId = randomUUID();
diff --git a/server/src/modules/agent-lifecycle/__tests__/diagnostics.test.ts b/server/src/modules/agent-lifecycle/__tests__/diagnostics.test.ts
index 4f7a8db7214..06b7e246d49 100644
--- a/server/src/modules/agent-lifecycle/__tests__/diagnostics.test.ts
+++ b/server/src/modules/agent-lifecycle/__tests__/diagnostics.test.ts
@@ -13,14 +13,58 @@ function fixture() {
const store = { get: vi.fn(), claim: vi.fn().mockResolvedValue(agent), renew: vi.fn().mockResolvedValue(true),
setRequiredPlugins: vi.fn().mockResolvedValue(true), completeHost: vi.fn().mockResolvedValue(true),
completePlugin: vi.fn().mockResolvedValue(true), defer: vi.fn().mockResolvedValue(true),
- change: vi.fn().mockResolvedValue({ ...agent, lifecycleState: "ready" }), pending: vi.fn().mockResolvedValue([agent.id]) } satisfies LifecycleStore;
+ change: vi.fn().mockResolvedValue({ ...agent, lifecycleState: "ready", lifecycleVersion: 4, lifecycleOperation: null }), pending: vi.fn().mockResolvedValue([agent.id]) } satisfies LifecycleStore;
const driver = { requiredPluginIds: vi.fn().mockResolvedValue(["plugin"]), runHost: vi.fn().mockResolvedValue("complete"),
- runPlugin: vi.fn().mockResolvedValue("complete") } satisfies LifecycleDriver;
+ runPlugin: vi.fn().mockResolvedValue("complete"), onReady: vi.fn().mockResolvedValue(undefined) } satisfies LifecycleDriver;
const report = vi.fn();
return { store, driver, report, worker: createLifecycleWorker(store, driver, report) };
}
afterEach(() => { vi.restoreAllMocks(); vi.useRealTimers(); });
+describe("lifecycle readiness handoff", () => {
+ it("notifies the scheduler only after host, plugins, and readiness have committed", async () => {
+ const { store, driver, worker } = fixture();
+ let commit!: (ready: LifecycleAgent) => void;
+ store.change.mockReturnValue(new Promise(resolve => { commit = resolve; }));
+ driver.onReady.mockImplementation(async ready => {
+ expect(store.completeHost).toHaveBeenCalledOnce();
+ expect(store.completePlugin).toHaveBeenCalledOnce();
+ expect(store.change).toHaveBeenCalledOnce();
+ expect(ready).toMatchObject({ lifecycleState: "ready", lifecycleVersion: 4, lifecycleOperation: null });
+ });
+ const work = worker.process(agent.id);
+ await vi.waitFor(() => expect(store.change).toHaveBeenCalledOnce());
+ expect(driver.onReady).not.toHaveBeenCalled();
+ commit({ ...agent, lifecycleState: "ready", lifecycleVersion: 4, lifecycleOperation: null });
+ await work;
+ expect(driver.onReady).toHaveBeenCalledOnce();
+ await worker.stop();
+ });
+
+ it.each(["host", "plugin", "transition", "hold"])("does not dispatch when %s prevents readiness", async stage => {
+ const { store, driver, worker } = fixture();
+ if (stage === "host") driver.runHost.mockResolvedValue("pending");
+ if (stage === "plugin") driver.runPlugin.mockResolvedValue("pending");
+ if (stage === "transition") store.change.mockResolvedValue(null);
+ if (stage === "hold") store.change.mockResolvedValue({ ...agent, lifecycleState: "paused" });
+ await worker.process(agent.id);
+ expect(driver.onReady).not.toHaveBeenCalled();
+ await worker.stop();
+ });
+
+ it("preserves committed readiness after a dispatch failure for the queue sweep to recover", async () => {
+ const { store, driver, report, worker } = fixture();
+ const error = new Error("queue unavailable");
+ driver.onReady.mockRejectedValue(error);
+ await worker.process(agent.id);
+ expect(report).toHaveBeenCalledWith(expect.objectContaining({ stage: "ready", phase: "ready", version: 4 }), error);
+ expect(store.defer).not.toHaveBeenCalled();
+ expect(driver.runHost).toHaveBeenCalledOnce();
+ expect(store.change).toHaveBeenCalledOnce();
+ await worker.stop();
+ });
+});
+
describe("lifecycle failure diagnostics", () => {
it.each(["host", "plugin", "claim", "renew", "scan", "defer"] as const)("reports %s failures with the available context", async stage => {
const { store, driver, report, worker } = fixture();
diff --git a/server/src/modules/agent-lifecycle/application/ports.ts b/server/src/modules/agent-lifecycle/application/ports.ts
index 1845ad8b7d1..11442b70b88 100644
--- a/server/src/modules/agent-lifecycle/application/ports.ts
+++ b/server/src/modules/agent-lifecycle/application/ports.ts
@@ -27,11 +27,13 @@ export interface LifecycleDriver {
requiredPluginIds(agent: LifecycleAgent): Promise;
runHost(agent: LifecycleAgent): Promise<"complete" | "pending">;
runPlugin(agent: LifecycleAgent, pluginId: string): Promise<"complete" | "pending">;
+ /** Notify the scheduler only after readiness has committed. */
+ onReady?(agent: LifecycleAgent): Promise;
}
export interface LifecycleFailure {
stage: "claim" | "renew" | "select_plugins" | "save_plugins" | "host" | "plugin"
- | "complete_host" | "complete_plugin" | "transition" | "defer" | "scan" | "policy_scan";
+ | "complete_host" | "complete_plugin" | "transition" | "defer" | "scan" | "policy_scan" | "ready";
agentId?: string;
companyId?: string;
phase?: AgentLifecycleState;
diff --git a/server/src/modules/agent-lifecycle/application/worker.ts b/server/src/modules/agent-lifecycle/application/worker.ts
index ad1c143dd96..3b90b227673 100644
--- a/server/src/modules/agent-lifecycle/application/worker.ts
+++ b/server/src/modules/agent-lifecycle/application/worker.ts
@@ -56,6 +56,14 @@ export function createLifecycleWorker(store: LifecycleStore, driver: LifecycleDr
if (!await runStep("complete_plugin", () => store.completePlugin(agent, owner, pluginId), pluginId)) continue phase;
}
const changed = await runStep("transition", () => store.change(id, "complete", { version: agent.lifecycleVersion, owner }));
+ if (changed?.lifecycleState === "ready" && !stopped && canRun()) {
+ // Accepted messages may already be waiting for verification. Do not
+ // leave them until the periodic queue sweep. Readiness is committed;
+ // a dispatch failure must not replay setup or defer the old phase.
+ await attempt({ ...context, stage: "ready", phase: changed.lifecycleState,
+ version: changed.lifecycleVersion, operationId: changed.lifecycleOperation?.id },
+ async () => { await driver.onReady?.(changed); }).catch(() => {});
+ }
if (changed && !isLifecycleWorkPending(changed.lifecycleState)) return;
} catch {
if (!await runStep("defer", () => store.defer(agent, owner, "The lifecycle step failed. Retry the operation.", new Date()))) continue;
diff --git a/server/src/services/agent-lifecycle-driver.ts b/server/src/services/agent-lifecycle-driver.ts
index 68cbea29999..11963d997fa 100644
--- a/server/src/services/agent-lifecycle-driver.ts
+++ b/server/src/services/agent-lifecycle-driver.ts
@@ -10,6 +10,9 @@ export function createLifecycleDriver(db: Db, manager: PluginWorkerManager): Lif
const { verify } = agentHarnessVerificationService(db, manager);
const heartbeat = heartbeatService(db, { pluginWorkerManager: manager });
return {
+ async onReady(agent) {
+ await heartbeat.resumeQueuedRunsForAgent(agent.companyId, agent.id);
+ },
async requiredPluginIds(agent) {
if (agent.lifecycleRequiredPluginIds !== null) return agent.lifecycleRequiredPluginIds;
const installed = await db.select().from(plugins).where(inArray(plugins.status, ["ready", "error", "installed", "upgrade_pending"]));
diff --git a/server/src/services/heartbeat.ts b/server/src/services/heartbeat.ts
index a549ee7ce42..961019eca34 100644
--- a/server/src/services/heartbeat.ts
+++ b/server/src/services/heartbeat.ts
@@ -8088,6 +8088,12 @@ export function heartbeatService(
resumeQueuedRuns,
+ resumeQueuedRunsForAgent: async (companyId: string, agentId: string) => {
+ const agent = await getAgent(agentId);
+ if (!agent || agent.companyId !== companyId) return [];
+ return startNextQueuedRunForAgent(agentId);
+ },
+
scheduleBoundedRetry,
reconcileStrandedAssignedIssues,