• 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.

• 🦀 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.

Best possible solution:

Land only after maintainers approve the unset-config default flip or revise the patch to keep upgrade compatibility while providing a clear opt-out/opt-in contract.

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

Yes, at source level. Current main reinjects default AGENTS.md sections when the config key is absent and keeps collapsed tool rows small, muted, nowrap, and ellipsized; I did not run a live browser capture in this read-only review.

Is this the best way to solve the issue?

Unclear as a final product choice. The implementation coherently reduces duplicate context and improves collapsed tool rows, but the unset-config default flip needs explicit maintainer approval or a compatibility-preserving path.

Label justifications:

• P2: The PR addresses bounded session-context and Control UI readability bugs with real upgrade impact but no core outage.
• merge-risk: 🚨 compatibility: The patch changes the meaning of an absent public config key for existing installations.
• merge-risk: 🚨 session-state: The changed compaction behavior affects which project instructions are carried into future agent session context.
• rating: 🦐 gold shrimp: Current PR rating is 🦐 gold shrimp because proof is 🦞 diamond lobster, patch quality is 🦐 gold shrimp, and The proof signal is strong, but patch readiness is limited by the compatibility-sensitive compaction default change that still needs maintainer approval.
• status: ⏳ waiting on author: ClawSweeper has contributor-facing work open and is waiting for author action. Sufficient (terminal): The PR body includes after-fix terminal verification plus browser computed-style proof for the Control UI CSS path, with the broad changed-gate gap disclosed.
• proof: sufficient: Contributor real behavior proof is sufficient. The PR body includes after-fix terminal verification plus browser computed-style proof for the Control UI CSS path, with the broad changed-gate gap disclosed.

Full review comments:

• [P1] Preserve unset-config upgrade behavior — src/auto-reply/reply/post-compaction-context.ts:98-99
  This changes absent agents.defaults.compaction.postCompactionSections from the current default reinjection path to no reinjection. Existing users who never set the key can silently lose post-compaction AGENTS.md guidance, so this needs explicit maintainer approval for the breaking policy or a compatibility-preserving opt-out/migration path before merge.
  Confidence: 0.88

Overall correctness: patch is incorrect
Overall confidence: 0.86

What I checked:

• Current main default behavior: Current main treats an absent agents.defaults.compaction.postCompactionSections as DEFAULT_POST_COMPACTION_SECTIONS, so unset configs still reinject the default AGENTS.md sections after compaction. (src/auto-reply/reply/post-compaction-context.ts:99, 5be62e779b2e)
• Current main safeguard suffix behavior: Current main's compaction safeguard appends workspace-critical-rules by calling readWorkspaceContextForSummary() in both provider and LLM suffix assembly paths without a config gate. (src/agents/pi-hooks/compaction-safeguard.ts:984, 5be62e779b2e)
• Documented current config contract: Current main docs and exported config comments say unset postCompactionSections defaults to ["Session Startup", "Red Lines"] and [] disables reinjection. Public docs: docs/gateway/config-agents.md. (docs/gateway/config-agents.md:681, 5be62e779b2e)
• PR behavior change: The PR diff changes the same path to return null unless postCompactionSections is an explicit non-empty array and passes that setting into the safeguard runtime. (src/auto-reply/reply/post-compaction-context.ts:98, f79b9619cf19)
• PR proof and status context: The PR body includes final-head format, lint, targeted Vitest, focused-suite, autoreview, and browser computed-style proof; the live PR metadata shows authorAssociation MEMBER and the protected maintainer label. (f79b9619cf19)
• Linked issue discussion: The linked session-bloat discussion explicitly suggested making section parsing opt-in, while the linked UI issue/comments identify the collapsed tool row's contrast, truncation, smaller text, and stray with wording as the user-visible problem.

Likely related people:

• efe-arv: Commit 03b9aba added configurable post-compaction context sections and the unset-config default contract that this PR changes. (role: introduced config behavior; confidence: high; commits: 03b9abab8486; files: src/auto-reply/reply/post-compaction-context.ts, src/config/types.agent-defaults.ts, src/config/schema.help.ts)
• BunsDev: Git history maps BunsDev to Val Alexander, who authored prior merged dashboard-v2/tool-card work on the affected Control UI rendering and CSS paths, beyond this PR. (role: Control UI feature-history owner; confidence: high; commits: f76a3c5225bb, ca414735b9ee, df72ca1ecec7; files: ui/src/styles/chat/tool-cards.css, ui/src/ui/chat/grouped-render.ts, ui/src/ui/chat/tool-cards.ts)
• 康熙: Commit 35a3e1b added the original post-compaction workspace context injection path that later became configurable. (role: introduced behavior; confidence: medium; commits: 35a3e1b788d6; files: src/auto-reply/reply/post-compaction-context.ts)
• Peter Steinberger: Current-main blame in this checkout points the active compaction and tool-card lines to a recent broad main snapshot commit, which is useful routing context but weak feature ownership proof. (role: recent line carrier; confidence: low; commits: b972ac194042; files: src/auto-reply/reply/post-compaction-context.ts, src/agents/pi-hooks/compaction-safeguard.ts, ui/src/styles/chat/tool-cards.css)

Codex review notes: model gpt-5.5, reasoning high; reviewed against 5be62e779b2e.

• 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.
• Hatchability usually comes from sufficient real-behavior proof, no blocking P0/P1/P2 findings, no security attention needed, and clean correctness. A merged PR is already final, so merge makes the egg hatchable independently.
• 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.