Best possible solution:

Land the narrow compat alignment after maintainer acceptance or a redacted live Xiaomi multi-turn tool replay proves the external API path.

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

Yes for the source-level payload path: current main only enables provider replay injection for DeepSeek, while the sibling agent transport path already enables it for Xiaomi. A live Xiaomi 400 was not reproduced in this review.

Is this the best way to solve the issue?

Yes, this is an acceptable narrow fix for the documented xiaomimimo.com endpoint because the changed provider detectCompat function owns the affected payload path. A shared helper could reduce future drift, but it is not necessary for this small alignment.

AGENTS.md: found and applied where relevant.

Codex review notes: model gpt-5.5, reasoning high; reviewed against 589ea28dabc8.

Label changes:

• add P2: This fixes a limited provider compatibility bug affecting Xiaomi/MiMo reasoning tool-history replay without broad core-runtime impact.
• add merge-risk: 🚨 compatibility: The diff changes default outbound payload behavior for existing Xiaomi or xiaomimimo.com OpenAI-compatible provider configurations.
• add rating: 🦐 gold shrimp: Overall readiness is 🦐 gold shrimp; proof is 🦪 silver shellfish and patch quality is 🐚 platinum hermit.
• add status: 📣 needs proof: The PR needs real behavior proof before ClawSweeper can clear the contributor ask. Needs real behavior proof before merge: The PR includes official-docs evidence and a unit payload capture, but no live Xiaomi request was run; redacted real behavior proof is still needed before merge unless maintainers override it. 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.
• remove rating: 🐚 platinum hermit: Current PR rating is rating: 🦐 gold shrimp, so this older rating label is no longer current.
• remove status: 👀 ready for maintainer look: Current PR status label is status: 📣 needs proof.

Label justifications:

• P2: This fixes a limited provider compatibility bug affecting Xiaomi/MiMo reasoning tool-history replay without broad core-runtime impact.
• merge-risk: 🚨 compatibility: The diff changes default outbound payload behavior for existing Xiaomi or xiaomimimo.com OpenAI-compatible provider configurations.
• rating: 🦐 gold shrimp: Overall readiness is 🦐 gold shrimp; proof is 🦪 silver shellfish and patch quality is 🐚 platinum hermit.
• status: 📣 needs proof: The PR needs real behavior proof before ClawSweeper can clear the contributor ask. Needs real behavior proof before merge: The PR includes official-docs evidence and a unit payload capture, but no live Xiaomi request was run; redacted real behavior proof is still needed before merge unless maintainers override it. 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.

PR surface:

Source +3, Tests +76. Total +79 across 2 files.

What I checked:

• Repository policy read: Root review policy was read fully; it makes provider routing and external API behavior compatibility-sensitive and requires external API proof when feasible. (AGENTS.md:15, 589ea28dabc8)
• Scoped agent guidance read: The only scoped AGENTS.md relevant to the sibling agent transport path is test-performance guidance and did not conflict with the review. (src/agents/AGENTS.md:1, 589ea28dabc8)
• Current provider path misses Xiaomi: Current main's provider-layer detectCompat sets requiresReasoningContentOnAssistantMessages from isDeepSeek only, so Xiaomi provider-layer replay does not get the field by default. (src/llm/providers/openai-completions.ts:1290, 589ea28dabc8)
• Replay injection call path: The provider payload builder only injects empty reasoning_content when the resolved compat flag is true and the model is reasoning-capable. (src/llm/providers/openai-completions.ts:1062, 589ea28dabc8)
• Sibling agent transport already covers Xiaomi: The agent transport compat defaults classify Xiaomi endpoints as DeepSeek-style and require reasoning_content replay, confirming the invariant the PR is aligning with. (src/agents/openai-completions-compat.ts:75, 589ea28dabc8)
• PR head changes provider compat defaults: The PR head adds isXiaomi for provider === "xiaomi" or xiaomimimo.com, then uses it for requiresReasoningContentOnAssistantMessages and thinkingFormat: "deepseek". (src/llm/providers/openai-completions.ts:1277, 50c79524037c)

Likely related people:

• @vincentkoc: Git blame in the available checkout attributes the central provider detectCompat function and sibling agent compat defaults to ed7f259ce7; history is flattened, so this is a routing signal rather than blame. (role: recent area contributor; confidence: medium; commits: ed7f259ce794; files: src/llm/providers/openai-completions.ts, src/agents/openai-completions-compat.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.