agents: (GPT 5.4 Parity vs. Opus Agentic) auto-activate strict-agentic for GPT-5 and emit blocked-exit liveness by 100yenadmin · Pull Request #64679 · openclaw/openclaw

100yenadmin · 2026-04-11T06:50:08Z

Summary

Two runtime fixes that close the remaining holes on completion-gate criteria 1 and 4 from #64227. Both are places where the first-wave parity work (PRs A #64241 and C #64300) did the right thing under an opt-in strict-agentic contract but left the default path short of the bar.

Criterion 1 ("GPT-5.4 no longer stalls after planning") is now universal. Out-of-the-box GPT-5 openai and openai-codex users get the 2-retry + blocked-state treatment automatically. Explicit executionContract: "default" still opts out.
Criterion 4 ("replay/liveness failures are surfaced as explicit states, not silent disappearance") no longer has a hole at the strict-agentic blocked exit. Every terminal exit in run.ts now sets replayInvalid + livenessState + calls setTerminalLifecycleMeta.

Stacked conceptually on PR A (#64241, merged) and PR C (#64300, merged); refines both without re-opening them. Part of #64227.

What changed

1. Default-contract stall blocker

src/agents/execution-contract.ts gets a new resolveEffectiveExecutionContract(params):

explicit contract	provider/model	effective
`"strict-agentic"`	supported (openai/openai-codex + gpt-5*)	`"strict-agentic"`
`"default"`	any	`"default"` (opt-out)
unspecified	supported	`"strict-agentic"` ← new
unspecified	unsupported	`"default"`
`"strict-agentic"`	unsupported	`"default"` (inert)

isStrictAgenticExecutionContractActive delegates to the resolver, so every caller — the update_plan auto-enable in openclaw-tools.registration.ts, the 2-retry cap in pi-embedded-runner/run.ts, the blocked-state branch in the same file — picks up the new semantics in lockstep.

Before: a user on openai/gpt-5.4 without executionContract set got 1 retry, then the code fell through to the normal success path and returned the plan-only assistant text as a completion. Criterion 1 was violated on the default install.

After: the same user gets the strict-agentic 2-retry cap and then STRICT_AGENTIC_BLOCKED_TEXT with isError: true. Criterion 1 is satisfied for the universal statement in #64227.

2. Blocked-exit liveness

src/agents/pi-embedded-runner/run.ts already threads replayInvalid + livenessState through every other early-return branch in runEmbeddedPiAgent, plus a attempt.setTerminalLifecycleMeta?.(...) call and the returned meta object. The strict-agentic blocked exit added in PR A #64241 didn't do any of that, so it was silent disappearance at exactly the point where the parity program said blocked state must be loudest.

This PR adds the terminal-state threading to that exit:

livenessState: "blocked". The EmbeddedRunLivenessState union is "working" | "paused" | "blocked" | "abandoned". "blocked" is the accurate label here because the run is waiting on an explicit act-now decision from the contract, not because the user walked away (which is what "abandoned" means in the same contract).
replayInvalid, delegated to the shared resolveReplayInvalidForAttempt(null) helper. The plan-only transcript is typically replay-safe even though the run is terminal, so the helper returns false, but routing through it keeps the replay-metadata edge cases consistent with the rest of the file.
attempt.setTerminalLifecycleMeta?.(...), using the same call pattern as every other terminal branch.

Regressions added

auto-enables update_plan for unconfigured GPT-5 openai runs
respects explicit default contract opt-out on GPT-5 runs
does not auto-enable update_plan for non-openai providers even when unconfigured
emits explicit replayInvalid + blocked liveness state at the strict-agentic blocked exit
auto-activates strict-agentic for unconfigured GPT-5 openai runs and surfaces the blocked state
respects explicit default contract opt-out on GPT-5 openai runs

Validation

pnpm test \
  src/agents/openclaw-tools.update-plan.test.ts \
  src/agents/pi-embedded-runner/run.incomplete-turn.test.ts \
  src/agents/pi-embedded-runner.buildembeddedsandboxinfo.test.ts \
  src/agents/system-prompt.test.ts \
  src/agents/openclaw-tools.sessions.test.ts \
  src/agents/pi-embedded-runner/run.overflow-compaction.test.ts

→ 122/122 passing. Rebased on green main (HEAD 545490c592 at push time).

Non-goals

doesn't change the EmbeddedRunLivenessState enum or add new states
doesn't touch the planning-only detector regexes or the gpt-5 model-prefix gate
doesn't touch PR D (benchmarks: add first-wave GPT-5.4 vs Opus 4.6 parity harness #64441) parity harness logic, PR E (test(qa): (GPT 5.4 Parity vs. Opus Agentic) expand GPT-5.4 / Opus parity second wave #64662) scenario expansion, or PR M (docs(parity): (GPT 5.4 Parity vs. Opus Agentic) catch up 10-PR program state with architecture diagrams #64837) documentation catch-up — those stay independent
doesn't restructure the default contract's retry-limit-exhausted fall-through for non-GPT-5 openai runs; they still get the legacy default behavior

greptile-apps · 2026-04-11T06:53:18Z

Greptile Summary

This PR closes two runtime contract gaps in the GPT-5.4 parity gate: (1) resolveEffectiveExecutionContract now auto-activates strict-agentic for unconfigured openai/openai-codex + gpt-5-family runs so the no-stall criterion applies out-of-the-box, not only for opt-in users; and (2) the strict-agentic blocked exit in run.ts now properly threads replayInvalid, livenessState: \"abandoned\", and setTerminalLifecycleMeta, consistent with every other terminal branch. Both changes are well-tested with targeted regressions and the logic is correct.

Confidence Score: 5/5

Safe to merge — changes are well-scoped, correctly implemented, and covered by new regressions.

All remaining findings are P2 (one minor test model-constant style issue per CLAUDE.md conventions). The core logic in execution-contract.ts and run.ts is correct, backwards-compatible, and the behavioral changes are intentional and explicitly documented.

No files require special attention.

Prompt To Fix All With AI

This is a comment left during a code review.
Path: src/agents/openclaw-tools.update-plan.test.ts
Line: 98

Comment:
**Non-preferred Anthropic model constant**

`CLAUDE.md` says to prefer `sonnet-4.6` for Anthropic model constants in tests, and to update such constants whenever the file is touched. This test was newly introduced in this PR, so it should use the preferred constant.

```suggestion
        modelId: "sonnet-4.6",
```

**Context Used:** CLAUDE.md ([source](https://app.greptile.com/review/custom-context?memory=fd949e91-5c3a-4ab5-90a1-cbe184fd6ce8))

How can I resolve this? If you propose a fix, please make it concise.

_{Reviews (1): Last reviewed commit: "agents: auto-activate strict-agentic for..." | Re-trigger Greptile}

Copilot

Pull request overview

This PR closes two runtime-contract gaps in the GPT‑5 parity work by (1) auto-activating the strict-agentic execution contract for supported GPT‑5 OpenAI/Codex runs when executionContract is unspecified (while preserving explicit "default" opt-out) and (2) ensuring the strict-agentic “planning-only retries exhausted” terminal exit emits replay/liveness metadata so observers don’t see a silent terminal transition.

Changes:

Add resolveEffectiveExecutionContract(...) and route strict-agentic activation checks through it to auto-enable strict-agentic for supported GPT‑5 OpenAI/Codex runs when unconfigured.
Thread replayInvalid + livenessState through the strict-agentic blocked-exit return path and attempt.setTerminalLifecycleMeta?.(...).
Add/expand regression tests covering auto-activation behavior, opt-out behavior, and terminal replay/liveness emission.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

File	Description
src/agents/pi-embedded-runner/run.ts	Adds replay/liveness terminal metadata emission for the strict-agentic blocked exit and types import for liveness state.
src/agents/pi-embedded-runner/run.incomplete-turn.test.ts	Adds regressions for strict-agentic blocked-exit liveness + GPT‑5 auto-activation and explicit default opt-out behavior.
src/agents/openclaw-tools.update-plan.test.ts	Updates gating tests to reflect strict-agentic auto-activation for unconfigured GPT‑5 OpenAI/Codex runs and explicit default opt-out.
src/agents/execution-contract.ts	Introduces an effective execution-contract resolver and updates strict-agentic activation logic to use it.

Copilot

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

Copilot

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated no new comments.

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: f811b4d773

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

Open a pull request for review
Mark a draft as ready
Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

…liveness Closes two hard blockers on the GPT-5.4 parity completion gate: 1) Criterion 1 (no stalls after planning) is universal, but the pre-existing strict-agentic execution contract was opt-in only. Out-of-the-box GPT-5 openai / openai-codex users who never set `agents.defaults.embeddedPi.executionContract` still got only 1 planning-only retry and then fell through to the normal completion path with the plan-only text, i.e. they still stalled. Introduce `resolveEffectiveExecutionContract(...)` in src/agents/execution-contract.ts. Behavior: - supported provider/model (openai or openai-codex + gpt-5-family) AND explicit "strict-agentic" or unspecified → "strict-agentic" - supported provider/model AND explicit "default" → "default" (opt-out) - unsupported provider/model → "default" regardless of explicit value `isStrictAgenticExecutionContractActive` now delegates to the effective resolver so the 2-retry + blocked-state treatment applies by default to every GPT-5 openai/codex run. Explicit opt-out still works for users who intentionally want the pre-parity-program behavior. 2) Criterion 4 (replay/liveness failures are explicit, not silent disappearance) is violated by the strict-agentic blocked exit itself. Every other terminal return path in src/agents/pi-embedded-runner/run.ts sets `replayInvalid` + `livenessState` via `setTerminalLifecycleMeta`, but the strict-agentic exit at run.ts:1615 falls through without them. Add explicit `livenessState: "abandoned"` + `replayInvalid` (via the shared `resolveReplayInvalidForAttempt` helper) to that exit, plus a `setTerminalLifecycleMeta` call so downstream observers (lifecycle log, ACP bridge, telemetry) see the same explicit terminal state they see on every other exit branch. Regressions added: - `auto-enables update_plan for unconfigured GPT-5 openai runs` - `respects explicit default contract opt-out on GPT-5 runs` - `does not auto-enable update_plan for non-openai providers even when unconfigured` - `emits explicit replayInvalid + abandoned liveness state at the strict-agentic blocked exit` - `auto-activates strict-agentic for unconfigured GPT-5 openai runs and surfaces the blocked state` - `respects explicit default contract opt-out on GPT-5 openai runs` Local validation: - pnpm test src/agents/openclaw-tools.update-plan.test.ts src/agents/pi-embedded-runner/run.incomplete-turn.test.ts src/agents/pi-embedded-runner.buildembeddedsandboxinfo.test.ts src/agents/system-prompt.test.ts src/agents/openclaw-tools.sessions.test.ts src/agents/pi-embedded-runner/run.overflow-compaction.test.ts 122/122 passing. Refs #64227

Triages all three loop-6 review comments on PR #64679: 1. Copilot: 'The strict-agentic blocked exit returns an error payload (isError: true) but sets livenessState to "abandoned". Elsewhere in the runner/lifecycle flow, error terminal states are treated as "blocked".' Verified: every other hardcoded error terminal branch in run.ts (role ordering at 1152, image size at 1206, schema error at 1244, compaction timeout at 1128, aborted-with-no-payloads at 606) uses livenessState: "blocked". Match that convention at the strict-agentic blocked exit at 1634. Updated the 'emits explicit replayInvalid + abandoned liveness state' regression test to assert the new "blocked" value and renamed the assertion commentary. 2. Copilot: 'The JSDoc for resolveEffectiveExecutionContract says explicit "strict-agentic" in config always resolves to "strict-agentic", but the implementation collapses to "default" whenever the provider/mode is unsupported.' Rewrite the JSDoc to explicitly document the unsupported-provider collapse as the lead case (strict-agentic is a GPT-5-family openai/openai-codex-only runtime contract) before listing the supported-lane behavior matrix. No code change; this is a docstring-only clarification. 3. Greptile P2: 'Non-preferred Anthropic model constant. CLAUDE.md says to prefer sonnet-4.6 for Anthropic test constants.' Swap claude-opus-4-6 → claude-sonnet-4-6 in the two update_plan gating fixtures that assert non-openai providers don't auto-enable the planning tool. Behavior unchanged; model constant now matches repo testing guidance. Local validation: - pnpm test src/agents/openclaw-tools.update-plan.test.ts src/agents/pi-embedded-runner/run.incomplete-turn.test.ts 29/29 passing. Refs #64227

… blocked state Addresses loop-7 Copilot finding on PR #64679: loop 6 changed the assertion to livenessState === 'blocked' to match the rest of the hard-error terminal branches in run.ts, but the test title still said 'abandoned liveness state', which made failures and test output misleading. Rename the test title to match the asserted value. No code change beyond the it(...) title. Validation: pnpm test src/agents/pi-embedded-runner/run.incomplete-turn.test.ts (19/19 pass). Refs #64227

…ariant GPT-5 model ids

100yenadmin · 2026-04-11T19:52:36Z

Current head cb35893753 is rebased onto latest main and the branch-owned runtime suite passed locally:

CI=1 pnpm exec vitest run src/agents/execution-contract.test.ts src/agents/pi-embedded-runner/run.incomplete-turn.test.ts

I also reran pnpm build after the rebase. The remaining build break reproduces on fresh main and is now isolated to shared blocker PR #64972 rather than this branch’s strict-agentic logic.

100yenadmin · 2026-04-11T23:43:12Z

Current head: cb35893753

Maintainer update after the fresh local verification pass:

Re-ran the branch-owned runtime suite on this head:
- CI=1 pnpm exec vitest run src/agents/execution-contract.test.ts src/agents/pi-embedded-runner/run.incomplete-turn.test.ts
- 34/34 passing
This is still the runtime slice that closes the last product gap behind the public claim:
- GPT-5-family openai / openai-codex runs auto-activate strict-agentic by default
- the strict-agentic blocked exit now emits explicit replay/liveness terminal metadata
Program-level proof update from the integrated patched stack:
- live GPT-5.4 passed the full 10-scenario agentic harness (10/10)
- after the qa-lab mock/provider follow-ups, the offline structural parity run also passes end-to-end (10/10 GPT lane, 10/10 Anthropic mock baseline, parity report pass)

I do not see remaining branch-owned review work here. If GitHub is still red, the remaining failures are the shared mainline type/check noise outside this diff, not this PR’s runtime scope.

100yenadmin · 2026-04-11T23:49:42Z

@steipete remaining GPT 5.4 fixes are ready for review. Built and completed live harness test (see PR for it + you can test as well if you like or use in future for tuning for models to openclaw by provider/model). Passes 10/10 parity scenario test side by side vs Opus 4.6 in Openclaw.

100yenadmin · 2026-04-12T06:56:49Z

Superseded by #65217. The runtime changes, tests, and current verification were carried forward into the runtime completion rollup so maintainers only need to review one runtime closeout PR. Closing this preserves the discussion history without splitting review attention.

* agents: auto-activate strict-agentic for GPT-5 and emit blocked-exit liveness Closes two hard blockers on the GPT-5.4 parity completion gate: 1) Criterion 1 (no stalls after planning) is universal, but the pre-existing strict-agentic execution contract was opt-in only. Out-of-the-box GPT-5 openai / openai-codex users who never set `agents.defaults.embeddedPi.executionContract` still got only 1 planning-only retry and then fell through to the normal completion path with the plan-only text, i.e. they still stalled. Introduce `resolveEffectiveExecutionContract(...)` in src/agents/execution-contract.ts. Behavior: - supported provider/model (openai or openai-codex + gpt-5-family) AND explicit "strict-agentic" or unspecified → "strict-agentic" - supported provider/model AND explicit "default" → "default" (opt-out) - unsupported provider/model → "default" regardless of explicit value `isStrictAgenticExecutionContractActive` now delegates to the effective resolver so the 2-retry + blocked-state treatment applies by default to every GPT-5 openai/codex run. Explicit opt-out still works for users who intentionally want the pre-parity-program behavior. 2) Criterion 4 (replay/liveness failures are explicit, not silent disappearance) is violated by the strict-agentic blocked exit itself. Every other terminal return path in src/agents/pi-embedded-runner/run.ts sets `replayInvalid` + `livenessState` via `setTerminalLifecycleMeta`, but the strict-agentic exit at run.ts:1615 falls through without them. Add explicit `livenessState: "abandoned"` + `replayInvalid` (via the shared `resolveReplayInvalidForAttempt` helper) to that exit, plus a `setTerminalLifecycleMeta` call so downstream observers (lifecycle log, ACP bridge, telemetry) see the same explicit terminal state they see on every other exit branch. Regressions added: - `auto-enables update_plan for unconfigured GPT-5 openai runs` - `respects explicit default contract opt-out on GPT-5 runs` - `does not auto-enable update_plan for non-openai providers even when unconfigured` - `emits explicit replayInvalid + abandoned liveness state at the strict-agentic blocked exit` - `auto-activates strict-agentic for unconfigured GPT-5 openai runs and surfaces the blocked state` - `respects explicit default contract opt-out on GPT-5 openai runs` Local validation: - pnpm test src/agents/openclaw-tools.update-plan.test.ts src/agents/pi-embedded-runner/run.incomplete-turn.test.ts src/agents/pi-embedded-runner.buildembeddedsandboxinfo.test.ts src/agents/system-prompt.test.ts src/agents/openclaw-tools.sessions.test.ts src/agents/pi-embedded-runner/run.overflow-compaction.test.ts 122/122 passing. Refs #64227 * agents: address loop-6 review comments on strict-agentic contract Triages all three loop-6 review comments on PR #64679: 1. Copilot: 'The strict-agentic blocked exit returns an error payload (isError: true) but sets livenessState to "abandoned". Elsewhere in the runner/lifecycle flow, error terminal states are treated as "blocked".' Verified: every other hardcoded error terminal branch in run.ts (role ordering at 1152, image size at 1206, schema error at 1244, compaction timeout at 1128, aborted-with-no-payloads at 606) uses livenessState: "blocked". Match that convention at the strict-agentic blocked exit at 1634. Updated the 'emits explicit replayInvalid + abandoned liveness state' regression test to assert the new "blocked" value and renamed the assertion commentary. 2. Copilot: 'The JSDoc for resolveEffectiveExecutionContract says explicit "strict-agentic" in config always resolves to "strict-agentic", but the implementation collapses to "default" whenever the provider/mode is unsupported.' Rewrite the JSDoc to explicitly document the unsupported-provider collapse as the lead case (strict-agentic is a GPT-5-family openai/openai-codex-only runtime contract) before listing the supported-lane behavior matrix. No code change; this is a docstring-only clarification. 3. Greptile P2: 'Non-preferred Anthropic model constant. CLAUDE.md says to prefer sonnet-4.6 for Anthropic test constants.' Swap claude-opus-4-6 → claude-sonnet-4-6 in the two update_plan gating fixtures that assert non-openai providers don't auto-enable the planning tool. Behavior unchanged; model constant now matches repo testing guidance. Local validation: - pnpm test src/agents/openclaw-tools.update-plan.test.ts src/agents/pi-embedded-runner/run.incomplete-turn.test.ts 29/29 passing. Refs #64227 * test: rename strict-agentic blocked-exit liveness regression to match blocked state Addresses loop-7 Copilot finding on PR #64679: loop 6 changed the assertion to livenessState === 'blocked' to match the rest of the hard-error terminal branches in run.ts, but the test title still said 'abandoned liveness state', which made failures and test output misleading. Rename the test title to match the asserted value. No code change beyond the it(...) title. Validation: pnpm test src/agents/pi-embedded-runner/run.incomplete-turn.test.ts (19/19 pass). Refs #64227 * agents: widen strict-agentic auto-activation to handle prefixed and variant GPT-5 model ids * Align strict-agentic retry matching * runtime: harden strict-agentic model matching --------- Co-authored-by: Eva <[email protected]>

* agents: auto-activate strict-agentic for GPT-5 and emit blocked-exit liveness Closes two hard blockers on the GPT-5.4 parity completion gate: 1) Criterion 1 (no stalls after planning) is universal, but the pre-existing strict-agentic execution contract was opt-in only. Out-of-the-box GPT-5 openai / openai-codex users who never set `agents.defaults.embeddedPi.executionContract` still got only 1 planning-only retry and then fell through to the normal completion path with the plan-only text, i.e. they still stalled. Introduce `resolveEffectiveExecutionContract(...)` in src/agents/execution-contract.ts. Behavior: - supported provider/model (openai or openai-codex + gpt-5-family) AND explicit "strict-agentic" or unspecified → "strict-agentic" - supported provider/model AND explicit "default" → "default" (opt-out) - unsupported provider/model → "default" regardless of explicit value `isStrictAgenticExecutionContractActive` now delegates to the effective resolver so the 2-retry + blocked-state treatment applies by default to every GPT-5 openai/codex run. Explicit opt-out still works for users who intentionally want the pre-parity-program behavior. 2) Criterion 4 (replay/liveness failures are explicit, not silent disappearance) is violated by the strict-agentic blocked exit itself. Every other terminal return path in src/agents/pi-embedded-runner/run.ts sets `replayInvalid` + `livenessState` via `setTerminalLifecycleMeta`, but the strict-agentic exit at run.ts:1615 falls through without them. Add explicit `livenessState: "abandoned"` + `replayInvalid` (via the shared `resolveReplayInvalidForAttempt` helper) to that exit, plus a `setTerminalLifecycleMeta` call so downstream observers (lifecycle log, ACP bridge, telemetry) see the same explicit terminal state they see on every other exit branch. Regressions added: - `auto-enables update_plan for unconfigured GPT-5 openai runs` - `respects explicit default contract opt-out on GPT-5 runs` - `does not auto-enable update_plan for non-openai providers even when unconfigured` - `emits explicit replayInvalid + abandoned liveness state at the strict-agentic blocked exit` - `auto-activates strict-agentic for unconfigured GPT-5 openai runs and surfaces the blocked state` - `respects explicit default contract opt-out on GPT-5 openai runs` Local validation: - pnpm test src/agents/openclaw-tools.update-plan.test.ts src/agents/pi-embedded-runner/run.incomplete-turn.test.ts src/agents/pi-embedded-runner.buildembeddedsandboxinfo.test.ts src/agents/system-prompt.test.ts src/agents/openclaw-tools.sessions.test.ts src/agents/pi-embedded-runner/run.overflow-compaction.test.ts 122/122 passing. Refs openclaw#64227 * agents: address loop-6 review comments on strict-agentic contract Triages all three loop-6 review comments on PR openclaw#64679: 1. Copilot: 'The strict-agentic blocked exit returns an error payload (isError: true) but sets livenessState to "abandoned". Elsewhere in the runner/lifecycle flow, error terminal states are treated as "blocked".' Verified: every other hardcoded error terminal branch in run.ts (role ordering at 1152, image size at 1206, schema error at 1244, compaction timeout at 1128, aborted-with-no-payloads at 606) uses livenessState: "blocked". Match that convention at the strict-agentic blocked exit at 1634. Updated the 'emits explicit replayInvalid + abandoned liveness state' regression test to assert the new "blocked" value and renamed the assertion commentary. 2. Copilot: 'The JSDoc for resolveEffectiveExecutionContract says explicit "strict-agentic" in config always resolves to "strict-agentic", but the implementation collapses to "default" whenever the provider/mode is unsupported.' Rewrite the JSDoc to explicitly document the unsupported-provider collapse as the lead case (strict-agentic is a GPT-5-family openai/openai-codex-only runtime contract) before listing the supported-lane behavior matrix. No code change; this is a docstring-only clarification. 3. Greptile P2: 'Non-preferred Anthropic model constant. CLAUDE.md says to prefer sonnet-4.6 for Anthropic test constants.' Swap claude-opus-4-6 → claude-sonnet-4-6 in the two update_plan gating fixtures that assert non-openai providers don't auto-enable the planning tool. Behavior unchanged; model constant now matches repo testing guidance. Local validation: - pnpm test src/agents/openclaw-tools.update-plan.test.ts src/agents/pi-embedded-runner/run.incomplete-turn.test.ts 29/29 passing. Refs openclaw#64227 * test: rename strict-agentic blocked-exit liveness regression to match blocked state Addresses loop-7 Copilot finding on PR openclaw#64679: loop 6 changed the assertion to livenessState === 'blocked' to match the rest of the hard-error terminal branches in run.ts, but the test title still said 'abandoned liveness state', which made failures and test output misleading. Rename the test title to match the asserted value. No code change beyond the it(...) title. Validation: pnpm test src/agents/pi-embedded-runner/run.incomplete-turn.test.ts (19/19 pass). Refs openclaw#64227 * agents: widen strict-agentic auto-activation to handle prefixed and variant GPT-5 model ids * Align strict-agentic retry matching * runtime: harden strict-agentic model matching --------- Co-authored-by: Eva <[email protected]>

Copilot AI review requested due to automatic review settings April 11, 2026 06:50

openclaw-barnacle Bot added agents Agent runtime and tooling size: M labels Apr 11, 2026

Copilot started reviewing on behalf of 100yenadmin April 11, 2026 06:50 View session

greptile-apps Bot reviewed Apr 11, 2026

View reviewed changes

Comment thread src/agents/openclaw-tools.update-plan.test.ts Outdated

Copilot AI reviewed Apr 11, 2026

View reviewed changes

Comment thread src/agents/pi-embedded-runner/run.ts Outdated

Comment thread src/agents/execution-contract.ts Outdated

100yenadmin changed the title ~~agents: auto-activate strict-agentic for GPT-5 and emit blocked-exit liveness~~ agents: (GPT 5.4 Parity vs. Opus Agentic) auto-activate strict-agentic for GPT-5 and emit blocked-exit liveness Apr 11, 2026

100yenadmin requested a review from Copilot April 11, 2026 13:57

Copilot started reviewing on behalf of 100yenadmin April 11, 2026 13:57 View session

Copilot AI reviewed Apr 11, 2026

View reviewed changes

Comment thread src/agents/pi-embedded-runner/run.incomplete-turn.test.ts Outdated

Comment thread src/agents/pi-embedded-runner/run.ts

100yenadmin force-pushed the fix/strict-agentic-default-plus-blocked-liveness branch from 41748de to 3913e9d Compare April 11, 2026 14:33

100yenadmin requested a review from Copilot April 11, 2026 15:00

Copilot started reviewing on behalf of 100yenadmin April 11, 2026 15:01 View session

Copilot AI reviewed Apr 11, 2026

View reviewed changes

100yenadmin force-pushed the fix/strict-agentic-default-plus-blocked-liveness branch from 3913e9d to f811b4d Compare April 11, 2026 16:49

openclaw-barnacle Bot added size: L and removed size: M labels Apr 11, 2026

chatgpt-codex-connector Bot reviewed Apr 11, 2026

View reviewed changes

Comment thread src/agents/execution-contract.ts Outdated

100yenadmin force-pushed the fix/strict-agentic-default-plus-blocked-liveness branch from 7b7a7da to 3df6f0b Compare April 11, 2026 18:41

Eva added 4 commits April 12, 2026 02:48

agents: widen strict-agentic auto-activation to handle prefixed and v…

1c0111f

…ariant GPT-5 model ids

Align strict-agentic retry matching

cb35893

100yenadmin force-pushed the fix/strict-agentic-default-plus-blocked-liveness branch from 3df6f0b to cb35893 Compare April 11, 2026 19:52

100yenadmin mentioned this pull request Apr 12, 2026

GPT-5.4 runtime completion rollup #65217

Closed

100yenadmin closed this Apr 12, 2026

100yenadmin mentioned this pull request Apr 12, 2026

agents: GPT-5.4 runtime completion rollup #65219

Merged

clawsweeper Bot mentioned this pull request May 9, 2026

fix(agent): detect CJK terminal continuation promises #79724

Closed

Uh oh!

Uh oh!

Conversation

100yenadmin commented Apr 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

What changed

1. Default-contract stall blocker

2. Blocked-exit liveness

Regressions added

Validation

Non-goals

Uh oh!

greptile-apps Bot commented Apr 11, 2026

Greptile Summary

Confidence Score: 5/5

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

chatgpt-codex-connector Bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

100yenadmin commented Apr 11, 2026

Uh oh!

100yenadmin commented Apr 11, 2026

Uh oh!

100yenadmin commented Apr 11, 2026

Uh oh!

100yenadmin commented Apr 12, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

100yenadmin commented Apr 11, 2026 •

edited

Loading