ClawSweeper status: review started.

I am starting a fresh review of this pull request: fix(llm): keep OpenAI-compatible reasoning streams active This is item 1/1 in the current shard. Shard 0/1.

This placeholder means the worker is alive and reading the current context. I will edit this same comment with the actual review when the claws are done clicking.

Crustacean status: shell secured, claws on keyboard, evidence pebbles being sorted.

Land-ready proof for exact head 31a3e181d2a05894bac03bbc97f7bf61dfb2ff84:

Local proof:

• node scripts/run-vitest.mjs src/agents/openai-transport-stream.test.ts src/agents/embedded-agent-runner/run/llm-idle-timeout.test.ts — 314 passed
• node ./node_modules/.bin/oxfmt --check src/agents/openai-transport-stream.ts src/agents/openai-transport-stream.test.ts && git diff --check — passed
• pnpm check:test-types — passed
• .agents/skills/autoreview/scripts/autoreview --mode local — clean, no accepted/actionable findings

CI proof:

• PR check rollup is green on the exact head, including CI selected shards, Real behavior proof, Security High, OpenGrep, dependency guard, workflow sanity, CodeQL critical-quality selection, and ci-timings-summary.
• The cancelled auto-response / Real behavior proof entries in the rollup were superseded by newer passing runs on the same head.

Known gap: no live reporter Vertex sidecar was available; regression models the reported OpenAI-compatible SSE shape with positive completion_tokens_details.reasoning_tokens in a usage-only chunk and verifies it emits an empty thinking_delta keepalive without adding trailing reasoning after visible text.