[nguyenxvotanminh3]

Summary

• Add STREAM_READINESS_MAX_TIMEOUT_MS so operators can raise the adaptive first-event cap for long-thinking agent workloads without changing code.
• Raise the default adaptive cap from 120s to 180s for very large/tool-heavy/high-reasoning streaming requests.
• Wire the cap into chat streaming readiness and document the env knob in .env.example.

Why

Large Codex/Kiro-style agent requests can legitimately take longer before producing the first useful SSE event. The existing adaptive policy could still cap at 120s, causing premature stream-readiness timeouts even though the stream may be alive.

Validation

• node --import tsx/esm --test tests/unit/runtime-timeouts.test.ts tests/unit/stream-readiness-policy.test.ts tests/unit/stream-readiness.test.ts tests/unit/combo-stream-readiness-fallback.test.ts tests/unit/chat-retry-policy.test.ts
• node --import tsx/esm --test tests/unit/runtime-timeouts.test.ts tests/unit/stream-readiness-policy.test.ts on the standalone branch
• npm run typecheck:core
• npx eslint src/shared/utils/runtimeTimeouts.ts open-sse/config/constants.ts open-sse/handlers/chatCore.ts open-sse/utils/streamReadinessPolicy.ts tests/unit/runtime-timeouts.test.ts tests/unit/stream-readiness-policy.test.ts
• git diff --check
• Runtime smoke: http://localhost:20129/api/auth/status returned 200 and resolver accepted STREAM_READINESS_MAX_TIMEOUT_MS=300000

[gemini-code-assist]

Warning

You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again!

[nguyenxvotanminh3]

Follow-up after review: verified the env override test is already asserting 300_000 for STREAM_READINESS_MAX_TIMEOUT_MS=300000 (no 30s typo in this branch). I also added the missing client-side protection for /v1/chat/completions: streaming requests now use the existing withEarlyStreamKeepalive wrapper while the handler is still waiting for upstream stream readiness, so strict clients receive SSE keepalives before the first real upstream event. Added a route-level regression test that delays the upstream stream past the 2s threshold and asserts the downstream body includes : omniroute-keepalive before the real SSE payload.\n\nValidation rerun on the PR branch:\n- node --import tsx/esm --test tests/unit/chat-combo-live-test.test.ts tests/unit/early-stream-keepalive.test.ts tests/unit/runtime-timeouts.test.ts tests/unit/stream-readiness-policy.test.ts (32/32)\n- git diff --check\n- npx eslint src/app/api/v1/chat/completions/route.ts tests/unit/chat-combo-live-test.test.ts (warnings only from pre-existing explicit-any lines in the test file)

[gemini-code-assist]

Warning

You have reached your daily quota limit. Please wait up to 24 hours and I will start processing your requests again!

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.

[diegosouzapw]

Thanks @nguyenxvotanminh3! The adaptive readiness cap + env knob + SSE keepalive solves real false-timeouts on high-reasoning runs. Merged into release/v3.8.43. 🙌