Add catch-all impact and merge-risk labels with required justification

[Takhoffman]

Summary

Add explicit catch-all labels for ClawSweeper-owned impact and merge-risk taxonomy gaps:

• impact:other
• merge-risk: 🚨 other

These labels should be real, searchable GitHub labels, but they should not carry the explanation themselves. The explanation must live in labelJustifications, just like the existing ClawSweeper-managed labels.

Problem

ClawSweeper currently supports a strict set of owned impact labels and merge-risk labels.

Impact labels are issue-only and currently limited to:

• impact:data-loss
• impact:security
• impact:crash-loop
• impact:message-loss
• impact:session-state
• impact:auth-provider

Merge-risk labels are PR-only and currently limited to:

• merge-risk: 🚨 compatibility
• merge-risk: 🚨 message-delivery
• merge-risk: 🚨 session-state
• merge-risk: 🚨 auth-provider
• merge-risk: 🚨 security-boundary
• merge-risk: 🚨 availability
• merge-risk: 🚨 automation

The current fallback for “no label applies” is [], which is correct for true absence of an owned signal. However, there is no searchable way to represent a real maintainer-visible impact or merge risk that does not fit the current taxonomy.

That makes two distinct cases look the same:

1. no meaningful owned impact or merge risk exists
2. meaningful impact or merge risk exists, but the taxonomy lacks a specific bucket

Those should not collapse into the same [] output.

Desired behavior

Add two catch-all labels:

impact:other

Use for issues only.

Meaning:

This issue has a meaningful maintainer-visible impact, but it does not fit the current owned impact taxonomy.

Rules:

• Use only for issues.
• Count toward the existing impactLabels max of 3.
• Prefer a specific impact label when one fits.
• Use [] when there is no meaningful owned impact signal.
• Require a matching labelJustifications entry explaining the actual impact.

merge-risk: 🚨 other

Use for pull requests only.

Meaning:

Merging this PR has a meaningful risk that green CI does not settle, but it does not fit the current owned merge-risk taxonomy.

Rules:

• Use only for PRs.
• Count toward the existing mergeRiskLabels max of 3.
• Prefer a specific merge-risk label when one fits.
• Use [] when there is no meaningful merge-risk signal.
• Require a matching labelJustifications entry explaining the actual risk.
• When selected, risks should describe the risk in maintainer-facing prose.
• When selected, mergeRiskOptions should still provide 1-3 concrete maintainer options.

Non-goals

• Do not add impact:n/a.
• Do not add merge-risk: n/a.
• Do not use other as a replacement for [].
• Do not allow other to bypass labelJustifications.
• Do not allow other to be preferred over a more specific existing label.

Implementation notes

Likely touchpoints:

• schema/clawsweeper-decision.schema.json

  • Add impact:other to impactLabels.items.enum.
  • Add merge-risk: 🚨 other to mergeRiskLabels.items.enum.
  • Add both labels to labelJustifications.items.properties.label.enum.
  • Update descriptions to define the catch-all behavior and clarify that other counts toward max 3.

• prompts/review-item.md

  • Add prompt guidance for both catch-all labels.
  • Make clear that specific labels win over other.
  • Make clear that [] still means no meaningful owned label applies.
  • Make clear that labelJustifications must explain the actual impact/risk when other is selected.
  • Make clear that merge-risk: 🚨 other still requires risks and mergeRiskOptions.

• Any label sync, generated label list, tests, fixtures, or docs if present.

Acceptance criteria

• ClawSweeper schema accepts impact:other.
• ClawSweeper schema accepts merge-risk: 🚨 other.
• labelJustifications accepts both new labels.
• Prompt guidance distinguishes:
  • [] = no meaningful owned label signal
  • other = meaningful signal exists, but no specific taxonomy bucket fits
• Prompt guidance says specific labels should be preferred over other.
• Prompt guidance says both other labels count toward the existing max of 3.
• Prompt guidance requires labelJustifications to explain the concrete impact or merge risk.
• Existing issue-only and PR-only boundaries remain:
  • impact labels are issue-only
  • merge-risk labels are PR-only

[Takhoffman]

@clawsweeper fix

[clawsweeper]

🦞🔧
ClawSweeper issue implementation requested.

I queued the repair worker to verify this issue on latest main and open or update one narrow implementation PR if it is safe.
Action: repair worker queued. Run: https://github.com/openclaw/clawsweeper/actions/runs/26374241280
Model: gpt-5.5

This lane creates PRs only; it does not merge or close the issue.

Result: implementation blocked.

• Reason: fix artifact is too broad for autonomous execution; split into narrower jobs or explicitly set CLAWSWEEPER_ALLOW_BROAD_FIX_ARTIFACTS=1
• Worker: https://github.com/openclaw/clawsweeper/actions/runs/26374241280
• Updated: 2026-05-24T22:18:06.685Z

[clawsweeper]

Codex review: keeping this open for maintainer follow-up; there is still a little grit to resolve. Reviewed May 24, 2026, 4:48 PM ET / 20:48 UTC.

Summary
Keep open: current main still has only the existing specific impact and merge-risk taxonomy, so the requested catch-all labels are not implemented and the issue is concrete enough for a narrow follow-up PR.

Reproducibility: not applicable. as a runtime reproduction: this is a feature request for label taxonomy behavior. Source inspection does confirm the requested labels are absent from the current schema, prompt, constants, and tests.

Ways to help us reproduce this

• Add a screenshot or short recording showing the behavior.
• Add expected vs actual behavior.
• Share version, platform, channel/provider, and relevant config details.

Next step
This is a bounded schema, prompt, label-sync, and test update with clear acceptance criteria and no linked implementation PR in the provided context.

Review details

Best possible solution:

Add the two catch-all labels consistently across the shared TypeScript label sets, schema, prompt guidance, label sync behavior, and tests while preserving specific-label preference and [] semantics.

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

Not applicable as a runtime reproduction: this is a feature request for label taxonomy behavior. Source inspection does confirm the requested labels are absent from the current schema, prompt, constants, and tests.

Is this the best way to solve the issue?

Yes, the requested direction is narrow and maintainable: update the existing centralized label taxonomy surfaces and add focused validation coverage rather than introducing a parallel label mechanism.

Codex review notes: model gpt-5.5, reasoning high; reviewed against 4954daa585f1.

Label changes

Label changes:

• add P3: This is a low-risk maintainer-triage taxonomy and prompt ergonomics improvement.
• add issue-rating: 🌊 off-meta tidepool: Current issue advisory state selects this label.
• add clawsweeper:queueable-fix: Current issue advisory state selects this label.
• add clawsweeper:fix-shape-clear: Current issue advisory state selects this label.

Label justifications:

• P3: This is a low-risk maintainer-triage taxonomy and prompt ergonomics improvement.

Evidence reviewed

Acceptance criteria:

• pnpm run test:unit
• pnpm run check

What I checked:

• Current schema excludes catch-all labels: The response schema enum for impactLabels, mergeRiskLabels, and labelJustifications[].label lists only the existing specific taxonomy labels and does not include impact:other or merge-risk: 🚨 other. (schema/clawsweeper-decision.schema.json:173, 4954daa585f1)
• Current TypeScript label sets exclude catch-all labels: The ImpactLabelName, MergeRiskLabelName, IMPACT_LABELS, and MERGE_RISK_LABELS definitions contain only the existing concrete label names, so parsing and label sync will not treat the requested catch-all labels as owned labels. (src/clawsweeper.ts:75, 4954daa585f1)
• Current prompt guidance preserves only [] fallback: The review prompt tells reviewers to use an empty array when no owned impact or merge-risk label applies, but it has no guidance distinguishing a true absence of signal from a meaningful signal outside the current taxonomy. (prompts/review-item.md:135, 4954daa585f1)
• Validation tests currently expect only specific labels: Unit tests assert the exact impact and merge-risk label schemes and invalid-label behavior against the current specific-label-only taxonomy, so the requested behavior needs test updates too. (test/clawsweeper.test.ts:10900, 4954daa585f1)
• History provenance for the taxonomy surface: Blame and pickaxe history point the existing impact/merge-risk taxonomy, schema, prompt, and labelJustifications surface back to commit 43cfca1, feat: fan out sweeps across target repositories. (src/clawsweeper.ts:75, 43cfca13e47e)

Likely related people:

• Peter Steinberger: Blame and pickaxe history show the existing impact/merge-risk label taxonomy, schema, prompt, tests, and labelJustifications surface were introduced in commit 43cfca1. (role: introduced behavior; confidence: high; commits: 43cfca13e47e; files: src/clawsweeper.ts, schema/clawsweeper-decision.schema.json, prompts/review-item.md)
• Tak Hoffman: Recent merged work in commit 4954daa refactored the PR review comment layout in src/clawsweeper.ts and test/clawsweeper.test.ts, adjacent to the public review output and label justification surfaces. (role: recent area contributor; confidence: medium; commits: 4954daa585f1; files: src/clawsweeper.ts, test/clawsweeper.test.ts)

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.

[Takhoffman]

@clawsweeper fix

[clawsweeper]

🦞👀
ClawSweeper picked this up.

Command router queued. I will update this comment with the next step.