[HAYDEN-OAI]

Summary

Make the basic ws example in docs/realtime.md distinguish a terminal Realtime response from a successful one.

Previously, the copied guide exited successfully after failed, cancelled, or incomplete responses, and after either a clean or abrupt socket close before response.done. All five failure cases reproduced against the unchanged guide and the real public SDK. A completed response remains the successful control.

The guide now uses the same small completion/status guards as the existing runnable Realtime examples: report an unsuccessful response or premature closure with a fixed diagnostic and a nonzero exit code, while preserving normal output and connection cleanup.

Scope and regression coverage

• Change only the handwritten guide and its existing executable-example test file; no SDK implementation, public types, dependencies, or generated files change.
• Execute the guide's exact first TypeScript fence through the public openai/realtime/ws entrypoint against a synthetic TLS/WebSocket loopback server.
• Cover completed, failed, cancelled, incomplete, clean early-close, and abrupt early-close cases, including the exact requests, synthetic authentication, output, and private-data omission.
• Preserve inherited runtime/proxy configuration and CA trust while isolating SDK credentials and endpoints in the fixture. Keep all 24 existing OpenAI/Azure example cases.

Validation

• Before the documentation fix: 5 new regressions failed, 1 success control passed; 24 existing cases were skipped for that targeted run.
• After: 30/30 focused tests passed on Node 22.22.3, 24.19.0, and 26.7.0.
• All 6 guide cases passed on exact Node 22.0.0.
• The exact guide fence passed strict public-declaration checks on TypeScript 4.9.5 and 6.0.3 (using compatible installed Node type definitions for each compiler).
• Canonical ./scripts/format, ./scripts/lint, and the normal repository TypeScript check passed.
• Full canonical handwritten suite: 7,737 tests passed across 203 files.
• Repeated adversarial review covered terminal status, premature closure, output draining, authentication transport differences, fixture cleanup, and supported-runtime compatibility.

[openai-sdks]

OkTest Summary

✅ 237/237 SDK tests passed in 10.736s for Node SDK PR #2647.

Test results — 42 files

Test	Result	Time
tests/chat-completions-complex-body.test.ts	✅ Passed	153ms
tests/chat-completions-create.test.ts	✅ Passed	218ms
tests/chat-completions-stream.test.ts	✅ Passed	253ms
tests/files-content-binary.test.ts	✅ Passed	130ms
tests/files-create-multipart.test.ts	✅ Passed	207ms
tests/files-list-pagination.test.ts	✅ Passed	293ms
tests/initialize-config.test.ts	✅ Passed	185ms
tests/instance-isolation.test.ts	✅ Passed	116ms
tests/models-list.test.ts	✅ Passed	185ms
tests/responses-background-lifecycle.test.ts	✅ Passed	192ms
tests/responses-body-method-errors.test.ts	✅ Passed	302ms
tests/responses-cancel-timeout.test.ts	✅ Passed	186ms
tests/responses-cancel.test.ts	✅ Passed	270ms
tests/responses-compact-retries.test.ts	✅ Passed	244ms
tests/responses-compact.test.ts	✅ Passed	216ms
tests/responses-create-advanced-stream.test.ts	✅ Passed	124ms
tests/responses-create-advanced.test.ts	✅ Passed	212ms
tests/responses-create-disconnect.test.ts	✅ Passed	1.13s
tests/responses-create-errors.test.ts	✅ Passed	191ms
tests/responses-create-malformed-api-responses.test.ts	✅ Passed	207ms
tests/responses-create-retries.test.ts	✅ Passed	257ms
tests/responses-create-stream-failures.test.ts	✅ Passed	262ms
tests/responses-create-stream-timeout.test.ts	✅ Passed	2.213s
tests/responses-create-stream-wire.test.ts	✅ Passed	2.664s
tests/responses-create-stream.test.ts	✅ Passed	90ms
tests/responses-create-terminal-states.test.ts	✅ Passed	206ms
tests/responses-create-timeout.test.ts	✅ Passed	213ms
tests/responses-create.test.ts	✅ Passed	209ms
tests/responses-delete.test.ts	✅ Passed	191ms
tests/responses-input-items-errors.test.ts	✅ Passed	323ms
tests/responses-input-items-list.test.ts	✅ Passed	273ms
tests/responses-input-items-options.test.ts	✅ Passed	206ms
tests/responses-input-tokens-count-timeout.test.ts	✅ Passed	228ms
tests/responses-input-tokens-count.test.ts	✅ Passed	228ms
tests/responses-malformed-inputs.test.ts	✅ Passed	2.237s
tests/responses-not-found-errors.test.ts	✅ Passed	226ms
tests/responses-parse.test.ts	✅ Passed	196ms
tests/responses-retrieve-retries.test.ts	✅ Passed	256ms
tests/responses-retrieve.test.ts	✅ Passed	214ms
tests/responses-stored-method-errors.test.ts	✅ Passed	563ms
tests/retry-behavior.test.ts	✅ Passed	3.038s
tests/sdk-error-shape.test.ts	✅ Passed	247ms

View OkTest run #33991657123

_{SDK merge (fc99a69e0697) · head (789fd044f4e0) · base (d0e2bcf021cb) · OkTest (2b1bdfd25e98)}

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-09-05T21:00:19.672690Z	789fd04	PR opened
🔒 Security Review	✅ Completed 2026-09-05T21:01:40.031811Z	789fd04	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[github-actions]

Castiron custom code

✅ No new custom-code files detected.

32 mixed files remain; 0 existing customizations changed.

Compared d0e2bcf021cb → 789fd044f4e0. Generated baselines verified.

32 existing customizations unchanged

• api.md
• scripts/castiron/README.md
• scripts/castiron/custom_code_report.py
• scripts/castiron/test_custom_code_report.py
• src/client.ts
• src/resources/audio/transcriptions.ts
• src/resources/audio/translations.ts
• src/resources/beta/assistants.ts
• src/resources/beta/beta.ts
• src/resources/beta/index.ts
• src/resources/beta/responses/internal-base.ts
• src/resources/beta/responses/responses.ts
• src/resources/beta/threads/index.ts
• src/resources/beta/threads/runs/index.ts
• src/resources/beta/threads/runs/runs.ts
• src/resources/beta/threads/threads.ts
• src/resources/chat/completions/completions.ts
• src/resources/chat/completions/index.ts
• src/resources/conversations/index.ts
• src/resources/embeddings.ts
• src/resources/files.ts
• src/resources/fine-tuning/checkpoints/permissions.ts
• src/resources/images.ts
• src/resources/responses/internal-base.ts
• src/resources/responses/responses.ts
• src/resources/skills/skills.ts
• src/resources/skills/versions/versions.ts
• src/resources/vector-stores/file-batches.ts
• src/resources/vector-stores/files.ts
• src/resources/webhooks/index.ts
• src/resources/webhooks/webhooks.ts
• tests/lib/data-residency.test.ts

A changed generated baseline means this report cannot reliably identify which handwritten lines changed.

Inspect the custom-code diff

Download the exact patch produced by this run (requires repository access):

gh run download 33991698708 --repo openai/openai-node \
  --name castiron-custom-code-33991698708-1 --dir /tmp/castiron-custom-code-33991698708-1
git apply --stat /tmp/castiron-custom-code-33991698708-1/custom-code.patch
cat /tmp/castiron-custom-code-33991698708-1/custom-code.patch

Or reproduce it from an SDK checkout containing the vendored reporter:

git fetch --no-tags origin d0e2bcf021cb73d8137179cf03948ff35e745f42 789fd044f4e0d273d0b6397e077616a544d905f2
python3 scripts/castiron/custom_code_report.py report \
  --base d0e2bcf021cb73d8137179cf03948ff35e745f42 \
  --head 789fd044f4e0d273d0b6397e077616a544d905f2 --fetch --require-head-hash --public \
  --out /tmp/castiron-custom-code-789fd044f4e0
cat /tmp/castiron-custom-code-789fd044f4e0/custom-code.patch

This is the current full custom patch for mixed files, not an attribution of only the handwritten lines changed by this PR.

Full report and patch

[markstuart-oai]

Reviewed the exact first TypeScript fence against the public Realtime WS implementation, response declarations, and all four runnable examples. The terminal-status check and single completion guard correctly distinguish unsuccessful responses and premature closure; fixed diagnostics avoid dumping response metadata or close reasons, and exitCode preserves output draining. The existing fixture is reused without unnecessary helpers or state, with synthetic credentials/endpoints and inherited runtime/proxy/CA configuration preserved. Scope stays within the two handwritten files; neither approaches 1,000 lines.

Independent validation: all six guide cases and twelve ws example cases passed on Node 22.22.3 and 24.20.0. Substituting the base guide reproduced five failures with the completed control passing. The package build, canonical lint, strict TypeScript 6.0.3 public-declaration check, and a 2 MiB output-draining probe through the built public entrypoint passed. Local limitations: twelve native websocket cases failed before reaching the fixture server, and the broad TypeScript check encountered missing optional example dependencies. I did not independently rerun the full suite, TypeScript 4.9, or exact Node 22.0.0. No substantive findings.

[jbeckwith-oai]

Reviewed the exact guide fence and executable regression fixture against the requested outcome. The response.done status guard correctly distinguishes completed responses from failed, cancelled, and incomplete terminal responses; the socket-close guard covers clean and abrupt premature closure without racing successful completion. Fixed diagnostics avoid leaking response metadata or close reasons, process.exitCode preserves output draining and cleanup, and the fixture exercises the public ws entrypoint with synthetic credentials across all six outcomes while preserving existing example coverage. Scope remains limited to the guide and its test, CI is green, and I found no in-scope blocking issues.