fix(agents): remove misleading run prefix from exec display for interpreter commands (fixes #97319) by zw-xysk · Pull Request #97326 · openclaw/openclaw

zw-xysk · 2026-06-27T20:47:42Z

What Problem This Solves

When an exec tool call fails, the cron run history and chat surface render the failure with a leading "run" that is a framework display label, not part of the command. This makes it look like the agent prepended "run" to the actual command.

Root Cause

summarizeKnownExec generates display summaries. For interpreter binaries (python3, node, ruby, php), it prepended "run" to the canonical bin+script form, producing e.g. "run python3 /path/to/script.py".

Changes Made

src/agents/tool-display-exec.ts: Changed interpreter fallback from run- <script> to just <script> so the display matches the actual command.
src/agents/tool-display.test.ts: Added regression test.

Evidence

Unit tests: 43/43 passed
Lint: clean
Lockfile: unchanged

AI Assistance

AI-assisted: Yes
Co-Authored-By: Claude [email protected]
Human confirmed understanding: Yes

…preter commands (openclaw#97319) For interpreter binaries (python3, node, ruby, php), the exec tool display summary showed "run python3 /path/to/script.py" which looks indistinguishable from the agent prepending "run" to the command. Drop the "run" prefix so the display shows the actual command as-is.

clawsweeper · 2026-06-27T20:48:51Z

Thanks for the context here. I swept through the related work, and this is now duplicate or superseded.

Close: this PR is superseded by the proof-positive shared failure-warning fix in #97511. This branch targets the same linked bug, still lacks real user-path proof, and its broad run-prefix stripper can degrade semantic exec summaries if merged as-is.

Root-cause cluster
Relationship: superseded
Canonical: #97511
Summary: This PR, the linked issue, and the overlapping PRs all target the exec failure-warning ambiguity; the proof-positive shared formatter PR is the viable canonical landing path.

Members:

canonical: fix: clarify exec tool failure warnings #97511 - It is open, mergeable, clean, proof-positive, and implements the shared payload-warning fix for the same linked bug.
fixed_by_candidate: [Bug]: exec tool failure message format makes framework labels indistinguishable from agent-typed commands #97319 - The canonical PR is a candidate fix for this issue, which describes the exact run python3 ... failure-warning ambiguity.
partial_overlap: fix(exec): keep framework verb out of backtick-wrapped failure message (#97319) #97425 - It targets the same issue but remains open with unknown mergeability and needs-proof labels, so it is not the safe canonical path.
superseded: fix(exec): keep framework verb out of backtick-wrapped failure message (#97319) #97359 - It was a closed, unmerged duplicate attempt for the same linked issue.

Proposal only: this assessment does not dispatch repair, suppress jobs, mutate sibling items, close, or merge anything.

Canonical path: Use #97511, or an equivalent maintainer-approved shared payload-warning formatter, as the landing path for the linked bug.

So I’m closing this here and keeping the remaining discussion on #97511.

Review details

Best possible solution:

Use #97511, or an equivalent maintainer-approved shared payload-warning formatter, as the landing path for the linked bug.

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

Yes. Current main still formats exec failure warnings as ⚠️ ${toolSummary} failed, and the exec display formatter can still produce run python3 ...; I did not run a live cron/chat flow in this read-only review.

Is this the best way to solve the issue?

No. The shared formatter in #97511 is the better fix because it separates the failure label from the command subject at the common payload-warning boundary without broadly stripping display metadata.

Security review:

Security review cleared: The diff only changes TypeScript presentation formatting and tests; it does not touch dependencies, workflows, permissions, secrets, install scripts, or package metadata.

AGENTS.md: found and applied where relevant.

What I checked:

Repository policy read: Root and scoped agent runner policies were read fully and applied; the relevant guidance required reviewing the shared warning path, caller/callee context, sibling surfaces, proof, and best-fix shape before verdict. (AGENTS.md:1, 245c18da5a6c)
Current main warning construction: Current main still builds visible tool failure text from formatToolAggregate(...) and then appends failed, so the warning boundary is the central shared surface for the reported ambiguity. (src/agents/embedded-agent-runner/run/payloads.ts:553, 245c18da5a6c)
Current main exec summary source: Current main summarizes Python and similar interpreter script invocations as run ${bin} ${script}, matching the linked issue's run python3 ... warning example. (src/agents/tool-display-exec.ts:240, 245c18da5a6c)
Current PR over-strips run metadata: At this PR head, stripLeadingExecDisplayVerb strips any multi-token metadata beginning with run, which can turn semantic summaries such as run sed on <file> into malformed command-looking warning text. (src/agents/tool-display-exec.ts:306, c992739b9667)
Superseding PR implementation: The canonical related PR implements a shared formatToolErrorWarningText path that separates the exec/bash tool label and failure state from the command subject instead of stripping metadata in place. (src/agents/embedded-agent-runner/run/payloads.ts:159, e170fa7a7876)
Superseding PR viability: The related PR is open, mergeable, clean, labeled proof: sufficient, and has no failed checks in the live status summary, so it is a viable canonical landing path rather than a closed or unproven replacement. (e170fa7a7876)

Likely related people:

Peter Steinberger: Introduced the split exec display parser and earlier extracted the compact tool-meta formatter that the warning path uses. (role: feature-history contributor; confidence: high; commits: ca0159569954, 597e7e6f1345; files: src/agents/tool-display-exec.ts, src/auto-reply/tool-meta.ts)
chenyangjun-xy: Authored recent merged work in buildEmbeddedRunPayloads, the surface that appends failure warnings from formatted tool metadata. (role: recent payload-warning contributor; confidence: medium; commits: e6f391234791; files: src/agents/embedded-agent-runner/run/payloads.ts)
Bartok: Authored recent merged work directly in the exec display formatter and adjacent tests. (role: recent exec-display contributor; confidence: medium; commits: 4d9cd7d22755; files: src/agents/tool-display-exec.ts, src/agents/tool-display.test.ts)
Vincent Koc: Git history shows recent current-main release/change history touching the exec display string history in this checkout, though the deeper feature ownership is shared with earlier parser and formatter work. (role: adjacent release/current-main contributor; confidence: low; commits: aa69b12d0086; files: src/agents/tool-display-exec.ts, src/auto-reply/tool-meta.ts)

Codex review notes: model internal, reasoning high; reviewed against 245c18da5a6c.

zw-xysk · 2026-06-27T20:59:28Z

@clawsweeper re-review

clawsweeper · 2026-06-27T20:59:32Z

🦞🧹
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.

harjothkhara · 2026-06-27T23:11:09Z

I did a source-level pass on the remaining scope here. The patch fixes the narrow interpreter-summary path in summarizeKnownExec, but the reported user-visible string is assembled one layer later: formatToolAggregate() still wraps exec/bash metadata as the command-shaped segment, and src/agents/embedded-agent-runner/run/payloads.ts still appends the failure suffix from that aggregate summary.

That makes the maintainer decision mostly about scope. If this PR is intended as the narrow python3 /path/to/script.py summary fix, the code shape looks aligned and the remaining proof should exercise a real cron/chat warning so reviewers can see the final rendered string. If it is intended to satisfy the broader wording in #97319 where the framework/tool label is outside the literal command, I would add a regression at the warning-payload or cron-diagnostics boundary rather than only formatToolDetail(), because that is the boundary that preserves or removes the ambiguous command-shaped failure text.

…n summaries (openclaw#97319) For exec/bash tool failures, the warning text embeds the exec display summary inside backticks before appending "failed". When the summary is a framework-injected "run <binary> <args>" pattern (e.g. "run python3 /path/to/script.py"), the leading "run" appears as if the agent typed it. Add stripLeadingExecDisplayVerb that collapses "run <bin> <args>" to "<bin> <args>" while preserving single-token map labels ("run tests", "run build") and non-run action verbs ("install dependencies", "show last 20 lines"). This complements the interpreter-summary fix in summarizeKnownExec by catching all exec/bash failure paths at the warning-rendering layer.

zw-xysk · 2026-06-28T18:45:50Z

Enhancement: strip `run` prefix from exec failure warnings

This adds a second layer to the fix: stripLeadingExecDisplayVerb in payloads.ts strips the framework-injected run verb from exec/bash failure warnings, complementing the summarizeKnownExec interpreter-summary fix already in this PR.

Changes

src/agents/tool-display-exec.ts — new stripLeadingExecDisplayVerb export: collapses run <bin> <args> → <bin> <args>, preserves single-token labels (run tests, run build) and non-run action verbs (install dependencies, show last 20 lines)
src/agents/embedded-agent-runner/run/payloads.ts — applies strip in the failure-warning path for exec/bash tools before formatToolAggregate wraps the meta
src/agents/tool-display.test.ts — 9 test cases covering strip, preserve, and edge cases

Effect

Before	After
⚠️ 🛠️ `run python3 /path` failed	⚠️ 🛠️ `python3 /path` failed
⚠️ 🛠️ `run tests` failed	⚠️ 🛠️ `run tests` failed (unchanged, single-token)
⚠️ 🛠️ `install dependencies` failed	⚠️ 🛠️ `install dependencies` failed (unchanged, not a `run` verb)

CI

136 checks passed. The single failure (checks-node-compact-large-whole-1) is an unrelated flaky session-lock test (attempt.session-lock.test.ts:3987), not related to this change.

@clawsweeper re-review

clawsweeper · 2026-06-28T18:45:53Z

🦞🧹
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.

vincentkoc · 2026-06-28T19:10:54Z

Closing this as superseded by the canonical shared formatting fix in #97511, landed as 3e90125. #97511 covers the broader payload warning boundary instead of only this narrower exec-summary slice. Thanks for pushing on the issue.

openclaw-barnacle Bot added agents Agent runtime and tooling size: XS labels Jun 27, 2026

This was referenced Jun 28, 2026

fix(exec): keep framework verb out of backtick-wrapped failure message (#97319) #97359

Closed

fix(exec): keep framework verb out of backtick-wrapped failure message (#97319) #97425

Closed

fix: clarify exec tool failure warnings #97511

Merged

openclaw-barnacle Bot added size: S and removed size: XS labels Jun 28, 2026

clawsweeper Bot added the merge-risk: 🚨 other 🚨 Merging this PR has meaningful risk outside the owned taxonomy. label Jun 28, 2026

vincentkoc closed this Jun 28, 2026

clawsweeper Bot mentioned this pull request Jul 6, 2026

fix(cron): summarize shell failures directly #92603

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Uh oh!

fix(agents): remove misleading run prefix from exec display for interpreter commands (fixes #97319)#97326

fix(agents): remove misleading run prefix from exec display for interpreter commands (fixes #97319)#97326
zw-xysk wants to merge 2 commits into
openclaw:mainfrom
zw-xysk:fix/issue-97319-exec-failure-label-format

zw-xysk commented Jun 27, 2026

Uh oh!

clawsweeper Bot commented Jun 27, 2026 •

edited

Loading

Uh oh!

zw-xysk commented Jun 27, 2026

Uh oh!

clawsweeper Bot commented Jun 27, 2026 •

edited

Loading

Uh oh!

harjothkhara commented Jun 27, 2026

Uh oh!

zw-xysk commented Jun 28, 2026

Uh oh!

clawsweeper Bot commented Jun 28, 2026 •

edited

Loading

Uh oh!

vincentkoc commented Jun 28, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Uh oh!

Conversation

zw-xysk commented Jun 27, 2026

What Problem This Solves

Root Cause

Changes Made

Evidence

AI Assistance

Uh oh!

clawsweeper Bot commented Jun 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

zw-xysk commented Jun 27, 2026

Uh oh!

clawsweeper Bot commented Jun 27, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

harjothkhara commented Jun 27, 2026

Uh oh!

zw-xysk commented Jun 28, 2026

Enhancement: strip run prefix from exec failure warnings

Changes

Effect

CI

Uh oh!

clawsweeper Bot commented Jun 28, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

vincentkoc commented Jun 28, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

clawsweeper Bot commented Jun 27, 2026 •

edited

Loading

clawsweeper Bot commented Jun 27, 2026 •

edited

Loading

Enhancement: strip `run` prefix from exec failure warnings

clawsweeper Bot commented Jun 28, 2026 •

edited

Loading