fix(agents): keep exec auto-reviewer rationale truncation UTF-16 safe by wm0018 · Pull Request #101513 · openclaw/openclaw

wm0018 · 2026-07-07T08:57:55Z

What Problem This Solves

Fixes an issue where exec auto-reviewer rationale and session prompt-template descriptions could render broken U+FFFD replacement characters (�) when the text contained an emoji or CJK supplementary character at the truncation boundaries (500 characters for rationale, 60 for descriptions).

String.prototype.slice cuts at UTF-16 code unit boundaries. Emoji like 🚀 are surrogate pairs (2 code units). When a pair straddles the cut position, the lone high surrogate renders as �.

Why This Change Was Made

Replace (text ?? fallback).slice(0, 500) in exec-auto-reviewer.ts and firstLine.slice(0, 60) in prompt-templates.ts with truncateUtf16Safe — the standard UTF-16-safe truncation helper already used across the codebase.

User Impact

Exec auto-reviewer rationale and session prompt-template descriptions containing emoji or CJK supplementary characters near the truncation limits now display cleanly truncated text, without broken replacement characters.

Evidence

Standalone proof (production helper, same head)

$ node --import tsx --input-type=module
node=v22.22.0
head=3ed27b43f9

// exec reviewer rationale (500 chars)
text.length=501
slice(0,500).lastCodeUnit=0xd83d
slice(0,500).hasLoneSurrogate=true
truncateUtf16Safe(text,500).matches-x499=true

// prompt template description (60 chars)
text.length=61
slice(0,60).lastCodeUnit=0xd83d
slice(0,60).hasLoneSurrogate=true
truncateUtf16Safe(text,60).matches-y59=true

Tests

$ node scripts/run-vitest.mjs src/agents/exec-auto-reviewer.test.ts

 Test Files  1 passed (1)
      Tests  15 passed (15)

Files Changed (3 files, +38/-3)

 src/agents/exec-auto-reviewer.test.ts       | 35 +++++++++
 src/agents/exec-auto-reviewer.ts            |  3 +-
 src/agents/sessions/prompt-templates.ts     |  4 +-

String.prototype.slice at offset 500 can split surrogate pairs in LLM-generated review rationale text, rendering broken U+FFFD in exec approval UI payloads. Replace raw slice(0, 500) with truncateUtf16Safe.

…tion truncation UTF-16 safe String.prototype.slice can split surrogate pairs at truncation boundaries, producing broken U+FFFD in both exec approval rationale (500 chars) and session prompt-template descriptions (60 chars). Replace raw slice with truncateUtf16Safe in both locations.

clawsweeper · 2026-07-07T09:17:55Z

Codex review: needs maintainer review before merge. Reviewed July 7, 2026, 6:40 AM ET / 10:40 UTC.

Summary
The PR changes exec auto-reviewer rationale normalization from .slice(0, 500) to truncateUtf16Safe(..., 500) and adds a regression test for a surrogate-pair boundary.

PR surface: Source +1, Tests +18. Total +19 across 2 files.

Reproducibility: yes. by source inspection: current main still truncates exec reviewer rationale with raw .slice(0, 500), which can split a surrogate pair at the boundary. I did not run tests in this read-only review.

Review metrics: none identified.

Stored data model
Persistent data-model change detected: serialized state: src/agents/exec-auto-reviewer.test.ts. Confirm migration or upgrade compatibility proof before merge.

Merge readiness
Overall: 🐚 platinum hermit
Proof: 🐚 platinum hermit
Patch quality: 🐚 platinum hermit
Result: ready for maintainer review.

Overall follows the weaker of proof and patch quality, so missing proof can cap an otherwise strong patch.

Rank-up moves:

[P2] Clear or rerun the failing checks-node-compact-small-whole-2 CI shard before merge.

Risk before merge

[P1] The live PR is behind the base branch and one broad CI shard is failing; I did not find evidence that the diff causes that failure, so this is merge handling rather than a code-review finding.

Maintainer options:

Decide the mitigation before merge
Land the focused helper-based exec reviewer fix after ordinary maintainer review and exact-head checks; handle other raw truncation sites as separate follow-up work if needed.
Pause or close
Do not merge this PR until maintainers decide whether the risk is worth taking.

Next step before merge

No repair lane is needed because there are no blocking review findings; remaining action is normal maintainer review plus CI/rebase handling.

Security
Cleared: The diff only changes TypeScript source and a focused test to use an existing internal helper; it does not touch dependencies, workflows, secrets, package metadata, generated assets, or code-execution setup.

Review details

Best possible solution:

Land the focused helper-based exec reviewer fix after ordinary maintainer review and exact-head checks; handle other raw truncation sites as separate follow-up work if needed.

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

Yes by source inspection: current main still truncates exec reviewer rationale with raw .slice(0, 500), which can split a surrogate pair at the boundary. I did not run tests in this read-only review.

Is this the best way to solve the issue?

Yes. Reusing the existing truncateUtf16Safe helper at normalizeRationale is the narrow owner-boundary fix; display-side cleanup or a broad raw-slice audit would be larger than this PR’s exec reviewer bug.

AGENTS.md: found and applied where relevant.

Codex review notes: model internal, reasoning high; reviewed against 942b44966107.

Label changes

Label justifications:

P3: The PR fixes rare text display corruption at one internal agent-review truncation boundary with limited blast radius.
rating: 🐚 platinum hermit: Overall readiness is 🐚 platinum hermit; proof is 🐚 platinum hermit and patch quality is 🐚 platinum hermit.
status: 👀 ready for maintainer look: ClawSweeper has no concrete contributor-facing blocker left for this PR. Sufficient (live_output): The PR body includes copied Node live output showing raw slice creates a lone surrogate and the existing helper returns clean truncated text; the current head keeps the same exec reviewer runtime change.
proof: sufficient: Contributor real behavior proof is sufficient. The PR body includes copied Node live output showing raw slice creates a lone surrogate and the existing helper returns clean truncated text; the current head keeps the same exec reviewer runtime change.

Evidence reviewed

PR surface:

Source +1, Tests +18. Total +19 across 2 files.

View PR surface stats

Area	Files	Added	Removed	Net
Source	1	2	1	+1
Tests	1	18	0	+18
Docs	0	0	0	0
Config	0	0	0	0
Generated	0	0	0	0
Other	0	0	0	0
Total	2	20	1	+19

What I checked:

Current main behavior: Current main still returns (text ?? fallback).slice(0, 500) from normalizeRationale, so a surrogate pair crossing the 500-code-unit boundary can be split before the decision rationale is returned. (src/agents/exec-auto-reviewer.ts:79, 942b44966107)
PR patch shape: The live PR file diff adds truncateUtf16Safe to src/agents/exec-auto-reviewer.ts and changes only the rationale truncation call plus one focused parser test. (src/agents/exec-auto-reviewer.ts:77, e97afe94d43f)
Helper contract: truncateUtf16Safe delegates to sliceUtf16Safe, which backs off when the truncation endpoint would leave a dangling surrogate half. (packages/normalization-core/src/utf16-slice.ts:44, 942b44966107)
Runtime entry point: createModelExecAutoReviewer is wired into the exec tool defaults, and node/gateway exec hosts call the reviewer before deciding whether to allow once or fall back to human approval. (src/agents/bash-tools.exec.ts:1488, 942b44966107)
Feature history: Blame in the available shallow history attributes the current normalizeRationale slice and shared UTF-16 helper to commit a6352f9356cc2ddf782059b66d042e2157e6d309; live PR metadata shows that commit came through merged pull request fix(status): surface auto-fallback model in status and session_status (#96126) #101337. (src/agents/exec-auto-reviewer.ts:77, a6352f9356cc)
Related search: Live GitHub searches found this PR as the only item directly owning the exec auto-reviewer rationale surrogate-boundary fix; older UTF-16 issues were adjacent helper or output surfaces, not this rationale path.

Likely related people:

LZY3538: Current-main blame in the available shallow history attributes both the affected normalizeRationale raw slice and the shared UTF-16 helper to a6352f9356cc2ddf782059b66d042e2157e6d309. (role: introduced behavior; confidence: medium; commits: a6352f9356cc; files: src/agents/exec-auto-reviewer.ts, packages/normalization-core/src/utf16-slice.ts)
steipete: Live GitHub metadata shows the introducing commit came through merged pull request fix(status): surface auto-fallback model in status and session_status (#96126) #101337, merged by steipete, and the current PR also has a maintainer follow-up test commit from steipete. (role: merger and recent follow-up owner; confidence: medium; commits: a6352f9356cc, e97afe94d43f; files: src/agents/exec-auto-reviewer.ts, src/agents/exec-auto-reviewer.test.ts, packages/normalization-core/src/utf16-slice.ts)

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.

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.

Review history (1 earlier review cycle)

reviewed 2026-07-07T09:21:11.250Z sha 3ed27b4 :: needs maintainer review before merge. :: none

steipete · 2026-07-07T11:04:05Z

Land-ready at exact reviewed head e97afe94d43f92c0413d0c0d9cd15d1ddf150bef.

Replaced raw rationale slicing in the exec auto-review response parser with the canonical normalization-core UTF-16 helper at the existing 500-code-unit cap.
Removed unrelated prompt-template churn and helper-only tests; the final regression asserts the actual parsed conservative decision shape.
Fresh Codex autoreview: clean, no accepted/actionable findings (correctness 0.93).
oxfmt on both touched files and git diff --check: passed.
Exact-head hosted CI: 71 checks passed. The only first-attempt failure was an unrelated TUI PTY timeout; workflow 28859095473 attempt 2 passed completely.
Native OPENCLAW_TESTBOX=1 scripts/pr prepare-run 101513: passed exact-head hosted gates.

No docs or changelog change is required for this internal parser-boundary correction.

steipete · 2026-07-07T11:04:23Z

Merged via squash.

Prepared head SHA: e97afe94d43f92c0413d0c0d9cd15d1ddf150bef
Landed commit: e7d617d4d9369a41d059f3345d4ec1e455ff635e

…openclaw#101513) * fix(agents): keep exec auto-reviewer rationale truncation UTF-16 safe String.prototype.slice at offset 500 can split surrogate pairs in LLM-generated review rationale text, rendering broken U+FFFD in exec approval UI payloads. Replace raw slice(0, 500) with truncateUtf16Safe. * fix(agents): keep exec reviewer rationale and prompt template description truncation UTF-16 safe String.prototype.slice can split surrogate pairs at truncation boundaries, producing broken U+FFFD in both exec approval rationale (500 chars) and session prompt-template descriptions (60 chars). Replace raw slice with truncateUtf16Safe in both locations. * test(exec): cover UTF-16-safe reviewer rationale --------- Co-authored-by: Peter Steinberger <steipete@gmail.com>

fix(agents): keep exec auto-reviewer rationale truncation UTF-16 safe

d38223c

String.prototype.slice at offset 500 can split surrogate pairs in LLM-generated review rationale text, rendering broken U+FFFD in exec approval UI payloads. Replace raw slice(0, 500) with truncateUtf16Safe.

openclaw-barnacle Bot added agents Agent runtime and tooling size: XS labels Jul 7, 2026

test(exec): cover UTF-16-safe reviewer rationale

e97afe9

steipete merged commit e7d617d into openclaw:main Jul 7, 2026
157 of 158 checks passed

github-actions Bot mentioned this pull request Jul 7, 2026

📡 Upstream Digest — 2026-07-07 14:10 UTC curtismercier/openclaw-mods#1230

Open

mushuiyu886 mentioned this pull request Jul 8, 2026

fix(agents): keep extension exec truncation UTF-16 safe #102007

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Uh oh!

fix(agents): keep exec auto-reviewer rationale truncation UTF-16 safe#101513

fix(agents): keep exec auto-reviewer rationale truncation UTF-16 safe#101513
steipete merged 3 commits into
openclaw:mainfrom
wm0018:fix/auto-reviewer-utf16-safe-truncation

wm0018 commented Jul 7, 2026 •

edited

Loading

Uh oh!

clawsweeper Bot commented Jul 7, 2026 •

edited

Loading

Uh oh!

steipete commented Jul 7, 2026

Uh oh!

Uh oh!

steipete commented Jul 7, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Uh oh!

Conversation

wm0018 commented Jul 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

What Problem This Solves

Why This Change Was Made

User Impact

Evidence

Standalone proof (production helper, same head)

Tests

Files Changed (3 files, +38/-3)

Uh oh!

clawsweeper Bot commented Jul 7, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

steipete commented Jul 7, 2026

Uh oh!

Uh oh!

steipete commented Jul 7, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

wm0018 commented Jul 7, 2026 •

edited

Loading

clawsweeper Bot commented Jul 7, 2026 •

edited

Loading