Best possible solution:

Land the narrow doctor-only debug override after normal maintainer review so routine doctor probes stay quiet while actionable runtime and sandbox policy audits remain visible.

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

Yes. Current main source shows removal audits emitted at info level from the shared policy pipeline, and the linked report gives a concrete doctor run with tools.profile set to coding; I did not rerun doctor because this review was read-only.

Is this the best way to solve the issue?

Yes. The branch is the best narrow fix because it adds an internal diagnostic log-level override only for doctor probe paths instead of globally lowering all policy-removal audits or suppressing actionable sandbox policy blocks.

AGENTS.md: found and applied where relevant.

Codex review notes: model gpt-5.5, reasoning high; reviewed against db7aff884361.

Label justifications:

• P2: This is a normal-priority doctor output bug fix with limited blast radius and focused regression coverage.
• rating: 🦞 diamond lobster: Overall readiness is 🦞 diamond lobster; proof is 🦞 diamond lobster and patch quality is 🦞 diamond lobster.
• status: 👀 ready for maintainer look: ClawSweeper has no concrete contributor-facing blocker left for this PR. Sufficient (terminal): The PR body includes after-fix Crabbox proof of a real doctor command path with exit 0 and an explicit assertion that the normal-level removal audit was absent.
• proof: sufficient: Contributor real behavior proof is sufficient. The PR body includes after-fix Crabbox proof of a real doctor command path with exit 0 and an explicit assertion that the normal-level removal audit was absent.

PR surface:

Source +17, Tests +40. Total +57 across 9 files.

What I checked:

• Repository policy read: Read the full root AGENTS.md and the scoped src/agents/AGENTS.md; the review applied the OpenClaw guidance for full PR review, scoped agents guidance, protected-label handling, and avoiding changelog findings. (AGENTS.md:1, db7aff884361)
• Current main behavior: Current main's auditToolPolicyFilter always emits removal audits through toolPolicyAuditLogger.info, which matches the linked issue's noisy normal-level doctor output. (src/agents/tool-policy-audit.ts:179, db7aff884361)
• PR implementation: The PR head adds ToolPolicyAuditLogLevel and routes removal-audit messages to debug only when the new logLevel parameter is set to debug; default behavior remains info. (src/agents/tool-policy-audit.ts:196, d7063e7d578b)
• Doctor-only wiring: The PR passes toolPolicyAuditLogLevel: "debug" from doctor runtime schema checks, while sibling runtime callers such as gateway tool resolution and embedded agent attempts still omit the override and keep info-level audits. (src/flows/doctor-core-checks.runtime.ts:686, d7063e7d578b)
• Regression coverage: The new pipeline test asserts debug is called and info is not called for diagnostic-only policy probes, while adjacent tests continue asserting ordinary allow/deny removals use info. (src/agents/tool-policy-pipeline.test.ts:327, d7063e7d578b)
• Logging contract: The logging subsystem defaults normal console output to info and only includes debug when verbose or debug logging is enabled, so moving these audits to debug keeps them out of normal doctor output. (src/logging/console.ts:29, db7aff884361)

Likely related people:

• Vincent Koc: Local history and blame show recent work on the tool-policy audit pipeline and doctor runtime schema checks that this PR adjusts. (role: recent area contributor; confidence: high; commits: edc0a22179a7, dc05f598bb5a; files: src/agents/tool-policy-audit.ts, src/agents/tool-policy-pipeline.ts, src/flows/doctor-core-checks.runtime.ts)
• Peter Steinberger: The timeline shows Peter assigned and force-pushed the current branch, and the final two PR commits scope the quieting to doctor paths. (role: recent PR follow-up owner; confidence: medium; commits: 276b158f364a, d7063e7d578b; files: src/flows/doctor-core-checks.runtime.ts, src/commands/doctor/shared/active-tool-schema-warnings.ts, src/agents/agent-tools.ts)

• 🦀 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.

• 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.