Best possible solution:

Choose one canonical Codex/OpenAI compaction-routing path that preserves existing session auth/runtime state, then land this narrow regression or fold its test coverage into the broader compaction-boundary fix.

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

Yes, by source and focused test path: current main's preflight compaction call uses the follow-up run route, while the PR constructs an over-threshold persisted Codex session and asserts the compaction call receives the persisted route. I did not run the tests in this read-only review.

Is this the best way to solve the issue?

Unclear as a final architecture choice: the patch is a narrow fix for current main's route mismatch, but sibling PRs may instead make Codex preflight compaction native-owned or handle broader OpenAI/Codex auth routing.

Codex review notes: model gpt-5.5, reasoning high; reviewed against ca70015a7ceb.

Label changes:

• add P1: The PR targets a broken agent/channel compaction route that can block Codex-backed Slack replies through auth-provider mismatch.
• add merge-risk: 🚨 session-state: The diff reuses persisted session runtime state during compaction, which must stay aligned with session ids and sibling compaction behavior.
• add merge-risk: 🚨 auth-provider: The diff changes which provider/auth profile budget preflight compaction uses for existing sessions.

Label justifications:

• P1: The PR targets a broken agent/channel compaction route that can block Codex-backed Slack replies through auth-provider mismatch.
• merge-risk: 🚨 auth-provider: The diff changes which provider/auth profile budget preflight compaction uses for existing sessions.
• merge-risk: 🚨 session-state: The diff reuses persisted session runtime state during compaction, which must stay aligned with session ids and sibling compaction behavior.
• rating: 🐚 platinum hermit: Overall readiness is 🐚 platinum hermit; proof is 🐚 platinum hermit and patch quality is 🦞 diamond lobster.
• status: 👀 ready for maintainer look: ClawSweeper has no concrete contributor-facing blocker left for this PR. Not applicable: This is a maintainer/member PR, so the contributor real-behavior-proof gate is not applied; the body provides focused mocked unit/e2e proof but no live Slack/auth-store run.

PR surface:

Source +27, Tests +133, Docs +1. Total +161 across 4 files.

What I checked:

• Protected maintainer item: The GitHub context marks the PR author association as MEMBER and includes the protected maintainer label, so cleanup closure is not allowed by the review policy.
• Current main still has the old preflight route: On current main, runPreflightCompactionIfNeeded passes compactEmbeddedPiSession the followup run provider/model and only the session agentHarnessId; it does not use the persisted session modelProvider/model/authProfileOverride route that this PR adds. (src/auto-reply/reply/agent-runner-memory.ts:756, ca70015a7ceb)
• Session store has the fields the PR preserves: SessionEntry persists authProfileOverride, modelProvider, model, and agentHarnessId, which are the route fields the proposed preflight helper forwards when the session id matches. (src/config/sessions/types.ts:308, ca70015a7ceb)
• Proposed diff is focused: The PR commit adds resolvePreflightCompactionRunTarget, forwards provider/model/authProfileId/agentHarnessId to compactEmbeddedPiSession, and adds focused unit plus runReplyAgent regression coverage for the Codex-backed Slack route. (src/auto-reply/reply/agent-runner-memory.ts, 66354a42582b)
• Sibling compaction work overlaps the route decision: Provided related-item context shows open PRs covering broader OpenAI compaction runtime routing, Codex compaction boundary behavior, and CLI runtime preflight gates, so landing order and intended boundary need maintainer coordination.
• History provenance: Blame ties the current preflight compaction call shape to recent main history, and file history shows repeated recent work on memory flush, preflight compaction, context windows, and embedded compaction routing. (src/auto-reply/reply/agent-runner-memory.ts:756, 5c7980fa1132)

Likely related people:

• Omar Shahine: git blame attributes the current preflight compaction call block in agent-runner-memory.ts to commit 5c7980f, which carried the current main shape being repaired. (role: introduced current preflight route shape; confidence: medium; commits: 5c7980fa1132; files: src/auto-reply/reply/agent-runner-memory.ts)
• Peter Steinberger: Recent history shows repeated work on memory flush policy, compaction runtime, embedded runner auth/runtime, and session routing surfaces adjacent to this PR. (role: recent area contributor; confidence: high; commits: a374c3a5bfd5, c90cb9c3c95c, 37625cff6fff; files: src/auto-reply/reply/agent-runner-memory.ts, src/agents/pi-embedded-runner/compact.ts)
• Jared: History shows the transcript-estimate preflight compaction behavior was added in commit c6d8318, making this person relevant to the threshold and preflight route path. (role: preflight compaction contributor; confidence: medium; commits: c6d8318d07f5; files: src/auto-reply/reply/agent-runner-memory.ts)
• Vincent Koc: History shows recent embedded compaction wrapper and provider/runtime seam work on compact.ts, which is the callee affected by the new preflight route parameters. (role: adjacent runtime contributor; confidence: medium; commits: f1c4e2f11daf, 74e7b8d47b18; files: src/agents/pi-embedded-runner/compact.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.

• 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.