codex: surface deferred dynamic tool names by steipete · Pull Request #83813 · openclaw/openclaw

steipete · 2026-05-18T23:50:34Z

Summary

add a compact developer-instruction manifest listing deferred searchable OpenClaw dynamic tool names
include the manifest for Codex app-server thread start/resume fallback paths and normal run attempts
cover deferred-tool and no-deferred-tool instruction rendering in thread lifecycle tests

Test plan

mise exec [email protected] -- pnpm vitest run extensions/codex/src/app-server/thread-lifecycle.test.ts
mise exec [email protected] -- pnpm exec tsc --noEmit --pretty false --project tsconfig.json

clawsweeper · 2026-05-18T23:51:49Z

Codex review: needs real behavior proof before merge.

Workflow note: Future ClawSweeper reviews update this same comment in place.

How this review workflow works

ClawSweeper keeps one durable marker-backed review comment per issue or PR.
Re-runs edit this comment so the latest verdict, findings, and automation markers stay together instead of adding duplicate bot comments.
A fresh review can be triggered by eligible @clawsweeper re-review comments, exact-item GitHub events, scheduled/background review runs, or manual workflow dispatch.
PR/issue authors and users with repository write access can comment @clawsweeper re-review or @clawsweeper re-run on an open PR or issue to request a fresh review only.
Maintainers can also comment @clawsweeper review to request a fresh review only.
Fresh-review commands do not start repair, autofix, rebase, CI repair, or automerge.
Maintainer-only repair and merge flows require explicit commands such as @clawsweeper autofix, @clawsweeper automerge, @clawsweeper fix ci, or @clawsweeper address review.
Maintainers can comment @clawsweeper explain to ask for more context, or @clawsweeper stop to stop active automation.

Summary
The PR adds a compact deferred dynamic-tool name manifest to Codex developer instructions, passes dynamic tool specs into Codex app-server prompt construction and snapshots, updates focused tests/fixtures, and includes a small Swift formatting cleanup.

Reproducibility: not applicable. this is a prompt-surface improvement rather than a bug report. Source inspection of the PR head shows the deferred-tool manifest path and focused test coverage.

PR rating
Overall: 🐚 platinum hermit
Proof: 🌊 off-meta tidepool
Patch quality: 🐚 platinum hermit
Summary: The patch looks normally mergeable after the loader repair, with focused tests and snapshot updates covering the changed prompt behavior.

Rank-up moves:

none

What the crustacean ranks mean

🦀 challenger crab: rare, exceptional readiness with strong proof, clean implementation, and convincing validation.
🦞 diamond lobster: very strong readiness with only minor maintainer review expected.
🐚 platinum hermit: good normal PR, likely mergeable with ordinary maintainer review.
🦐 gold shrimp: useful signal, but proof or patch confidence is still limited.
🦪 silver shellfish: thin signal; proof, validation, or implementation needs work.
🧂 unranked krab: not merge-ready because proof is missing/unusable or there are serious correctness or safety concerns.
🌊 off-meta tidepool: rating does not apply to this item.

Shiny media proof means a screenshot, video, or linked artifact directly shows the changed behavior. Runtime, network, CSP, and security claims still need visible diagnostics.

PR egg
✨ Hatched: 🥚 common Tiny Review Wisp

        .--^^^^--.           
     .-'  o    o  '-.        
    /       \__/      \      
   |    /\  ____  /\   |     
   |   /  \/____\/  \  |     
    \  \_.------._/  /       
     '._  `----'  _.'        
        '-.____.-'           
       _/|_|  |_|\_          
      /__|      |__\         
       .-----------.         
      '-------------'

Rarity: 🥚 common.
Trait: hums during re-review.
Share on X: post this hatch
Copy: My PR egg hatched a 🥚 common Tiny Review Wisp in ClawSweeper.

What is this egg doing here?

Eggs appear after the PR passes real-behavior proof. It is here for vibes, not verdicts: it does not change labels, ratings, merge decisions, or automation.
The shell reacts to review momentum: open follow-up work warms it up, re-review makes it wobble, and a clean final review lets it hatch.
How to hatch it: reach status: 👀 ready for maintainer look or status: 🚀 automerge armed; that usually means sufficient real-behavior proof, no blocking P0/P1/P2 findings, no security attention needed, and clean correctness.
The hatch is seeded from this repository and PR number, so the same PR keeps the same creature; the reviewed head SHA can only change safe visual details.
Rarity is just collectible sparkle: 🥚 common, 🌱 uncommon, 💎 rare, ✨ glimmer, and 🌈 legendary.

Real behavior proof
Not applicable: This protected maintainer-labeled Codex prompt/test-helper PR is maintainer-handled; the PR discussion lists focused local validation after the loader repair.

Next step before merge
No repair lane is needed; the prior loader blocker is already patched on the PR head and the remaining action is ordinary maintainer/CI handling.

Security
Cleared: No concrete security or supply-chain concern was found; the diff does not add dependencies, workflows, lifecycle scripts, credentials handling, or new downloaded code.

Review details

Best possible solution:

Keep the PR’s source-backed loader repair, let focused CI/maintainer checks confirm the prompt snapshot path, and merge only if maintainers accept the model-facing instruction wording and token cost.

Do we have a high-confidence way to reproduce the issue?

Not applicable; this is a prompt-surface improvement rather than a bug report. Source inspection of the PR head shows the deferred-tool manifest path and focused test coverage.

Is this the best way to solve the issue?

Yes. The current approach derives a compact manifest from existing deferLoading tool specs and uses the existing alias-aware source loader for prompt snapshots instead of adding a second loader path.

Label justifications:

P3: This is a low-blast-radius Codex prompt and test-helper improvement rather than an urgent user-facing regression.

What I checked:

Deferred manifest implementation: The PR head builds developer instructions with an optional dynamic-tools argument and inserts a sorted, de-duplicated list of tools whose specs have deferLoading set. (extensions/codex/src/app-server/thread-lifecycle.ts:829, 3147dc0ef7aa)
Start/resume propagation: Thread start and resume parameter builders now include dynamic tools when constructing default developer instructions, and run-attempt passes toolBridge specs into the base developer instructions. (extensions/codex/src/app-server/thread-lifecycle.ts:589, 3147dc0ef7aa)
Focused prompt tests: The PR adds tests proving deferred tools are surfaced, sorted, and omitted when no tool is deferred. (extensions/codex/src/app-server/thread-lifecycle.test.ts:71, 3147dc0ef7aa)
Loader finding repaired: The updated helper loads bundled plugin public source through getCachedPluginSourceModuleLoader with pluginSdkResolution set to src instead of using a bare Jiti loader. (src/test-utils/bundled-plugin-public-surface.ts:135, 3147dc0ef7aa)
Alias-aware loader contract: Current main's source module loader wrapper forces tryNative false, which routes source-backed plugin loads through the Jiti alias map used by plugin SDK resolution. (src/plugins/plugin-module-loader-cache.ts:329, 6f18decb7a2c)
Discussion validation: The PR discussion says the loader finding was addressed and lists prompt snapshot, focused Vitest, oxfmt, swiftformat, and git diff checks run after the repair. (3147dc0ef7aa)

Likely related people:

Peter Steinberger: History shows Peter introduced and repeatedly maintained the Codex app-server control/lifecycle surface touched by this PR. (role: Codex app-server feature-history owner; confidence: high; commits: 31a0b7bd42a5, 3b65e2302a55, b4fdd1470b70; files: extensions/codex/src/app-server/run-attempt.ts, extensions/codex/src/app-server/thread-lifecycle.ts)
Vincent Koc: Recent main history shows Vincent maintained nearby Codex runtime/auth behavior that shares the app-server lifecycle path. (role: recent adjacent Codex runtime contributor; confidence: medium; commits: 859eb0666282, f1cc8f0cfc7c, 4a4f52b09730; files: extensions/codex/src/app-server/run-attempt.ts, extensions/codex/src/app-server/thread-lifecycle.ts)
pashpashpash: The PR history shows pashpashpash authored the commits that made prompt snapshots source-backed through the alias-aware plugin loader and reported validation. (role: PR follow-up repair author; confidence: medium; commits: 3b4ab43a3603, 3147dc0ef7aa; files: src/test-utils/bundled-plugin-public-surface.ts, test/helpers/agents/happy-path-prompt-snapshots.ts)

Codex review notes: model gpt-5.5, reasoning high; reviewed against 6f18decb7a2c.

pashpashpash · 2026-05-19T00:10:08Z

@clawsweeper review

clawsweeper · 2026-05-19T00:11:42Z

🦞🧹
ClawSweeper re-review requested.

I asked ClawSweeper to review this item again.
Action: item re-review queued (workflow sweep.yml, event repository_dispatch).
Result: the existing ClawSweeper review comment will be edited in place when the review finishes.

Re-review progress:

State: Complete
Detail: The targeted re-review finished, the durable review comment was updated, and the synced verdict was routed.
Run: https://github.com/openclaw/clawsweeper/actions/runs/26068005662
Updated: 2026-05-19T00:17:26.086Z

pashpashpash · 2026-05-19T01:09:12Z

Addressed the ClawSweeper loader finding on the latest head.

The prompt snapshot helper no longer creates a bare jiti loader. It now calls a source-backed bundled-plugin public-surface helper, and that helper uses getCachedPluginSourceModuleLoader with pluginSdkResolution: "src", so Codex test-api.ts is loaded through the same alias-aware plugin loader path.

Validated locally:

node --import tsx scripts/generate-prompt-snapshots.ts --check
node scripts/run-vitest.mjs extensions/codex/src/app-server/thread-lifecycle.test.ts test/scripts/prompt-snapshots.test.ts -- --reporter=dot
./node_modules/.bin/oxfmt --check --threads=1 ...
swiftformat --lint apps/macos/Sources/OpenClaw/VoiceWakeSettings.swift --config config/swiftformat
git diff --check

@clawsweeper review

pashpashpash · 2026-05-19T01:10:02Z

@clawsweeper review

clawsweeper · 2026-05-19T01:10:54Z

🦞🧹
ClawSweeper re-review requested.

I asked ClawSweeper to review this item again.
Action: item re-review queued (workflow sweep.yml, event repository_dispatch).
Result: the existing ClawSweeper review comment will be edited in place when the review finishes.

Re-review progress:

State: Failed
Detail: The targeted re-review did not finish cleanly. Check the workflow run for details.
Run: https://github.com/openclaw/clawsweeper/actions/runs/26070223783
Updated: 2026-05-19T01:19:24.217Z

pashpashpash · 2026-05-19T01:19:33Z

Pushed a narrow follow-up for the loader repair on 118f3ddac3f85fd8cb183a4bc35afe4b909e3900.

The source-backed public-surface helper is now a properly typed generic function, with the same narrow oxlint suppression used by the existing test-loader generic surface. This fixes the exact CI failures from the prior head: check-prod-types, check-test-types, and check-lint.

Validated locally:

pnpm tsgo:core && pnpm tsgo:extensions
pnpm check:test-types
pnpm lint
node --import tsx scripts/generate-prompt-snapshots.ts --check
node scripts/run-vitest.mjs extensions/codex/src/app-server/thread-lifecycle.test.ts test/scripts/prompt-snapshots.test.ts -- --reporter=dot
./node_modules/.bin/oxfmt --check --threads=1 ...
swiftformat --lint apps/macos/Sources/OpenClaw/VoiceWakeSettings.swift --config config/swiftformat
git diff --check

I also tried pnpm check:changed; all JS/TS guards, type, and lint lanes passed, but the local machine cannot run SwiftLint because Homebrew requires full Xcode and the portable SwiftLint binary cannot load sourcekitdInProc here. The fresh PR CI is running the macOS lane in the proper environment.

@clawsweeper review

clawsweeper · 2026-05-19T01:21:17Z

🦞🧹
ClawSweeper re-review requested.

I asked ClawSweeper to review this item again.
Action: item re-review queued (workflow sweep.yml, event repository_dispatch).
Result: the existing ClawSweeper review comment will be edited in place when the review finishes.

Re-review progress:

State: Superseded
Detail: A newer re-review for this item started before this run finished, so GitHub cancelled this older run. Check the latest ClawSweeper run for the current result.
Run: https://github.com/openclaw/clawsweeper/actions/runs/26070323940
Updated: 2026-05-19T01:25:53.032Z

pashpashpash · 2026-05-19T01:25:31Z

Pushed one more narrow cleanup on 1f03376b8c07dcb209b002d4cceab1e934775e247 for the latest exact-head build-artifacts failure.

The failure was test/scripts/lint-suppressions.test.ts: my previous follow-up added a third no-unnecessary-type-parameters suppression in src/test-utils/bundled-plugin-public-surface.ts, while the allowlist expected two. The source loader did not actually need to be generic because the caller already casts the loaded Codex test API, so I removed the extra suppression and made the helper return object.

Validated locally:

node scripts/run-vitest.mjs test/scripts/lint-suppressions.test.ts -- --reporter=dot
pnpm check:test-types
pnpm lint
node --import tsx scripts/generate-prompt-snapshots.ts --check
./node_modules/.bin/oxfmt --check --threads=1 src/test-utils/bundled-plugin-public-surface.ts
git diff --check

@clawsweeper review

clawsweeper · 2026-05-19T01:27:21Z

🦞🧹
ClawSweeper re-review requested.

I asked ClawSweeper to review this item again.
Action: item re-review queued (workflow sweep.yml, event repository_dispatch).
Result: the existing ClawSweeper review comment will be edited in place when the review finishes.

Re-review progress:

State: Complete
Detail: The targeted re-review finished, the durable review comment was updated, and the synced verdict was routed.
Run: https://github.com/openclaw/clawsweeper/actions/runs/26070524752
Updated: 2026-05-19T01:33:51.215Z

* codex: surface deferred dynamic tool names * codex: keep prompt snapshots source-backed * style: wrap mac voice settings help text * style: satisfy swiftformat for voice wake help text * style: apply swiftformat to voice wake help text * test: load codex prompt snapshots through plugin aliases * test: type codex source surface loader * test: avoid extra codex loader suppression --------- Co-authored-by: pashpashpash <[email protected]>

openclaw-barnacle Bot added extensions: codex size: S maintainer Maintainer-authored PR labels May 18, 2026

clawsweeper Bot added rating: 🐚 platinum hermit Good normal PR readiness with ordinary maintainer review expected. status: 👀 ready for maintainer look ClawSweeper has no concrete contributor-facing blocker left for this PR. labels May 18, 2026

steipete force-pushed the codex-deferred-tool-manifest branch from 26137be to 48cf481 Compare May 19, 2026 00:05

steipete and others added 3 commits May 18, 2026 17:49

codex: surface deferred dynamic tool names

3b1323f

codex: keep prompt snapshots source-backed

3b4ab43

style: wrap mac voice settings help text

b56dc8b

pashpashpash force-pushed the codex-deferred-tool-manifest branch from 7485057 to b56dc8b Compare May 19, 2026 00:51

openclaw-barnacle Bot added the app: macos App: macos label May 19, 2026

pashpashpash added 2 commits May 18, 2026 17:54

style: satisfy swiftformat for voice wake help text

cc60b3c

style: apply swiftformat to voice wake help text

d6deb2e

clawsweeper Bot added rating: 🦐 gold shrimp Decent PR readiness signal, but merge confidence is limited. and removed rating: 🦪 silver shellfish Thin PR readiness signal; proof, validation, or implementation needs work. labels May 19, 2026

test: load codex prompt snapshots through plugin aliases

3147dc0

clawsweeper Bot added rating: 🐚 platinum hermit Good normal PR readiness with ordinary maintainer review expected. and removed rating: 🦐 gold shrimp Decent PR readiness signal, but merge confidence is limited. labels May 19, 2026

test: type codex source surface loader

118f3dd

test: avoid extra codex loader suppression

1f03376

pashpashpash merged commit 0b4fc26 into main May 19, 2026
99 checks passed

pashpashpash deleted the codex-deferred-tool-manifest branch May 19, 2026 01:32

Haderach-Ram mentioned this pull request May 19, 2026

Ecosystem Digest — 2026-05-19 Haderach-Ram/openclaw-radar#12

Open

github-actions Bot mentioned this pull request May 19, 2026

📡 Upstream Digest — 2026-05-19 02:33 UTC curtismercier/openclaw-mods#895

Open

clawsweeper Bot mentioned this pull request Jun 14, 2026

feat: add tool search directory mode #91632

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Uh oh!

codex: surface deferred dynamic tool names#83813

codex: surface deferred dynamic tool names#83813
pashpashpash merged 8 commits into
mainfrom
codex-deferred-tool-manifest

steipete commented May 18, 2026

Uh oh!

clawsweeper Bot commented May 18, 2026 •

edited

Loading

Uh oh!

pashpashpash commented May 19, 2026

Uh oh!

clawsweeper Bot commented May 19, 2026 •

edited

Loading

Uh oh!

pashpashpash commented May 19, 2026

Uh oh!

pashpashpash commented May 19, 2026

Uh oh!

clawsweeper Bot commented May 19, 2026 •

edited

Loading

Uh oh!

pashpashpash commented May 19, 2026

Uh oh!

clawsweeper Bot commented May 19, 2026 •

edited

Loading

Uh oh!

pashpashpash commented May 19, 2026

Uh oh!

clawsweeper Bot commented May 19, 2026 •

edited

Loading

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Uh oh!

Conversation

steipete commented May 18, 2026

Summary

Test plan

Uh oh!

clawsweeper Bot commented May 18, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pashpashpash commented May 19, 2026

Uh oh!

clawsweeper Bot commented May 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pashpashpash commented May 19, 2026

Uh oh!

pashpashpash commented May 19, 2026

Uh oh!

clawsweeper Bot commented May 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pashpashpash commented May 19, 2026

Uh oh!

clawsweeper Bot commented May 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pashpashpash commented May 19, 2026

Uh oh!

clawsweeper Bot commented May 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

clawsweeper Bot commented May 18, 2026 •

edited

Loading

clawsweeper Bot commented May 19, 2026 •

edited

Loading

clawsweeper Bot commented May 19, 2026 •

edited

Loading

clawsweeper Bot commented May 19, 2026 •

edited

Loading

clawsweeper Bot commented May 19, 2026 •

edited

Loading