• 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:

Rebase on current main, keep the custom-provider JSON/HTML diagnostic changes, restore the node-host Codex routing fix and its tests/changelog, then add redacted real OpenClaw/custom-provider proof or get an explicit maintainer proof override.

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

Yes. Source inspection shows current main accepts any 2xx custom-provider verification response and the OpenAI SDK posts Chat Completions to /chat/completions, while the PR head clearly removes the current-main node-host Codex routing guard and tests.

Is this the best way to solve the issue?

No for the current branch. The custom-provider diagnostic approach is narrow in isolation, but the force-pushed Codex/changelog regression must be removed before this is a safe fix path.

Label changes:

• add P1: The current PR head can regress a recently fixed command execution routing path for users relying on Codex app-server with exec host=node.
• add merge-risk: 🚨 compatibility: The force-pushed branch removes current-main behavior that preserves existing host=node exec routing setups.
• add merge-risk: 🚨 security-boundary: Dropping the node-host guard can run shell-capable native Code Mode on the gateway rather than through OpenClaw's selected node execution path.
• add merge-risk: 🚨 availability: Node-targeted Codex runs can lose the OpenClaw exec/process tools needed to complete remote-node command workflows.
• add rating: 🧂 unranked krab: Current PR rating is 🧂 unranked krab because proof is 🦪 silver shellfish, patch quality is 🧂 unranked krab, and The provider fix has focused tests, but mock-only proof plus an unrelated Codex execution regression make the current PR not quality-ready.
• remove P2: Current review triage priority is P1, so this older priority label is no longer current.
• remove rating: 🦪 silver shellfish: Current PR rating is rating: 🧂 unranked krab, so this older rating label is no longer current.

Label justifications:

• P1: The current PR head can regress a recently fixed command execution routing path for users relying on Codex app-server with exec host=node.
• merge-risk: 🚨 compatibility: The force-pushed branch removes current-main behavior that preserves existing host=node exec routing setups.
• merge-risk: 🚨 security-boundary: Dropping the node-host guard can run shell-capable native Code Mode on the gateway rather than through OpenClaw's selected node execution path.
• merge-risk: 🚨 availability: Node-targeted Codex runs can lose the OpenClaw exec/process tools needed to complete remote-node command workflows.
• rating: 🧂 unranked krab: Current PR rating is 🧂 unranked krab because proof is 🦪 silver shellfish, patch quality is 🧂 unranked krab, and The provider fix has focused tests, but mock-only proof plus an unrelated Codex execution regression make the current PR not quality-ready.
• status: 📣 needs proof: The PR needs real behavior proof before ClawSweeper can clear the contributor ask. Needs real behavior proof before merge: The PR body reports loopback Vitest coverage only; it needs redacted real OpenClaw/custom-provider terminal output, logs, screenshot, recording, or linked artifact after the patch before merge unless maintainers override proof. After adding proof, update the PR body; ClawSweeper should re-review automatically. If it does not, the PR author or someone with repository write access can comment @clawsweeper re-review.

Full review comments:

• [P1] Restore node-host Codex tool routing — extensions/codex/src/app-server/run-attempt.ts:3669-3675
  Current main added isEffectiveExecHostNode and addNodeShellDynamicToolsIfNeeded so /exec host=node disables native Codex Code Mode and still exposes OpenClaw exec/process. This branch removes that path, so node-targeted Codex runs can execute on the gateway/native app-server surface instead of the selected node and lose the tools that route through OpenClaw. Please rebase and keep the current-main node-host routing behavior and tests.
  Confidence: 0.93

Overall correctness: patch is incorrect
Overall confidence: 0.91

Security concerns:

• [high] Node-host shell routing guard removed — extensions/codex/src/app-server/run-attempt.ts:3669
  By removing the host=node native-surface guard and OpenClaw exec/process reinjection, the PR can route shell-capable Codex turns to the wrong execution boundary.
  Confidence: 0.9

What I checked:

• Current main custom-provider verification behavior: Current main still treats any 2xx custom-provider verification probe as successful, which matches the linked report that an HTML web-root response can pass setup without a useful baseUrl hint. (src/commands/onboard-custom.ts:110, 5cc0dbce8628)
• OpenAI SDK endpoint contract: The OpenAI Node SDK Chat Completions resource posts to /chat/completions, so OpenAI-compatible providers served under /v1 need that path in the configured baseURL.
• PR custom-provider implementation: The branch adds successful-response JSON validation for onboarding probes and a runtime OpenAI-compatible HTML-response hint that includes a redacted configured baseUrl. (src/commands/onboard-custom.ts:95, 0bba7f652d68)
• Regression introduced by current PR head: The PR head removes current-main node-host handling by calling only addSandboxShellDynamicToolsIfAvailable, so node-targeted Codex runs no longer re-add OpenClaw exec and process after native Code Mode is disabled. (extensions/codex/src/app-server/run-attempt.ts:3669, 0bba7f652d68)
• Current-main node-host fix provenance: Current main commit 5cc0dbc added isEffectiveExecHostNode, addNodeShellDynamicToolsIfNeeded, regression tests, and a changelog entry for routing /exec host=node through OpenClaw tools instead of native Codex Code Mode. (extensions/codex/src/app-server/run-attempt.ts:3730, 5cc0dbce8628)
• Diff sanity check: The diff is whitespace-clean, but it includes unrelated Codex and changelog changes beyond the custom-provider diagnostic fix. (0bba7f652d68)

Likely related people:

• vincentkoc: Authored current-main commit 5cc0dbc that added the Codex node-host routing behavior this PR head now removes. (role: recent area contributor; confidence: high; commits: 5cc0dbce8628; files: extensions/codex/src/app-server/run-attempt.ts, extensions/codex/src/app-server/run-attempt.test.ts, CHANGELOG.md)
• steipete: Blame on the custom-provider onboarding, guarded fetch, and surrounding Codex app-server code in this shallow checkout points mostly to Peter Steinberger's recent baseline commit. (role: adjacent owner; confidence: medium; commits: f4bdfd46a99a; files: src/commands/onboard-custom.ts, src/agents/provider-transport-fetch.ts, src/agents/openai-transport-stream.ts)
• openperf: Git history for node-host execution includes prior fixes around exec host=node routing, making this a useful routing candidate if maintainers need more context. (role: related historical contributor; confidence: medium; commits: d98eaba4c3b3, 77509024b8c6; files: extensions/codex/src/app-server/run-attempt.ts)

Codex review notes: model gpt-5.5, reasoning high; reviewed against 5cc0dbce8628.

• The egg game starts only after the PR passes the real-behavior proof check.
• Before that, no creature or rarity is rolled. The treat waits for real proof.
• This is still just collectible flavor: proof affects review readiness, not creature quality.