[HAYDEN-OAI]

Security findings

• HIGH csf_60ad55d472f1de23a525c274: unauthenticated assistant listing exposes account data.
• MEDIUM csf_97c0b7e3dc03a6c21566a620: anonymous completion and streaming routes proxy paid model requests.
• MEDIUM csf_c2b9fc853edab3a40da1eb0f: anonymous transcription requests consume paid capacity.
• MEDIUM csf_3d634c846334bb5128ffb403 and csf_e2bb95e5859e20ee92db380c: public Edge/Node test routes incur charges and leave uploaded files behind.

Changes

• Require a separate, fail-closed bearer token for every /api/* route before OpenAI client initialization; reject oversized requests and bound chat inputs/model output.
• Generate cryptographically random credentials for existing CI smoke runs and allow browser users to enter the token without exposing it in public build configuration.
• Delete every uploaded ecosystem-test file in finally, including assertion-failure paths, across both Edge and Node runtimes.
• Add focused authentication, request-boundary, and file-cleanup regressions; restrict every change to ecosystem-tests/vercel-edge/**.

Verification

• Five focused Jest suites: 55 passing tests; 20 cleanup regressions failed before the fix.
• Ecosystem TypeScript check: node_modules/.bin/tsc --noEmit --incremental false.
• Production application build: npm run build.
• Full repository lint with pinned Oxfmt 0.62.0: ./scripts/lint.
• Real development/production HTTP checks: seven anonymous routes and spoofed middleware headers return 401; oversized authorized input returns 413; authorized requests reach the existing handler; the public homepage remains 200. No paid upstream requests were made.

Rollout

Deployments must configure a private VERCEL_EDGE_TEST_TOKEN containing at least 32 characters; without it, API routes remain safely disabled. Existing ecosystem CI generates a fresh token automatically. Public SDK APIs and generated/upstream-owned files are unchanged.

[openai-sdks]

OkTest Summary

✅ 237/237 SDK tests passed in 10.291s for Node SDK PR #2412.

Test results — 42 files

Test	Result	Time
tests/chat-completions-complex-body.test.ts	✅ Passed	145ms
tests/chat-completions-create.test.ts	✅ Passed	212ms
tests/chat-completions-stream.test.ts	✅ Passed	208ms
tests/files-content-binary.test.ts	✅ Passed	155ms
tests/files-create-multipart.test.ts	✅ Passed	142ms
tests/files-list-pagination.test.ts	✅ Passed	146ms
tests/initialize-config.test.ts	✅ Passed	158ms
tests/instance-isolation.test.ts	✅ Passed	109ms
tests/models-list.test.ts	✅ Passed	144ms
tests/responses-background-lifecycle.test.ts	✅ Passed	263ms
tests/responses-body-method-errors.test.ts	✅ Passed	258ms
tests/responses-cancel-timeout.test.ts	✅ Passed	221ms
tests/responses-cancel.test.ts	✅ Passed	218ms
tests/responses-compact-retries.test.ts	✅ Passed	265ms
tests/responses-compact.test.ts	✅ Passed	188ms
tests/responses-create-advanced-stream.test.ts	✅ Passed	188ms
tests/responses-create-advanced.test.ts	✅ Passed	159ms
tests/responses-create-disconnect.test.ts	✅ Passed	1.177s
tests/responses-create-errors.test.ts	✅ Passed	260ms
tests/responses-create-malformed-api-responses.test.ts	✅ Passed	120ms
tests/responses-create-retries.test.ts	✅ Passed	266ms
tests/responses-create-stream-failures.test.ts	✅ Passed	231ms
tests/responses-create-stream-timeout.test.ts	✅ Passed	2.168s
tests/responses-create-stream-wire.test.ts	✅ Passed	2.477s
tests/responses-create-stream.test.ts	✅ Passed	121ms
tests/responses-create-terminal-states.test.ts	✅ Passed	289ms
tests/responses-create-timeout.test.ts	✅ Passed	224ms
tests/responses-create.test.ts	✅ Passed	174ms
tests/responses-delete.test.ts	✅ Passed	254ms
tests/responses-input-items-errors.test.ts	✅ Passed	249ms
tests/responses-input-items-list.test.ts	✅ Passed	173ms
tests/responses-input-items-options.test.ts	✅ Passed	121ms
tests/responses-input-tokens-count-timeout.test.ts	✅ Passed	227ms
tests/responses-input-tokens-count.test.ts	✅ Passed	209ms
tests/responses-malformed-inputs.test.ts	✅ Passed	2.143s
tests/responses-not-found-errors.test.ts	✅ Passed	262ms
tests/responses-parse.test.ts	✅ Passed	199ms
tests/responses-retrieve-retries.test.ts	✅ Passed	225ms
tests/responses-retrieve.test.ts	✅ Passed	209ms
tests/responses-stored-method-errors.test.ts	✅ Passed	501ms
tests/retry-behavior.test.ts	✅ Passed	2.812s
tests/sdk-error-shape.test.ts	✅ Passed	305ms

View OkTest run #32199256039

_{SDK merge (756dbc848ba1) · head (bf3aa24e0365) · base (b7a0bd6144ed) · OkTest (7e7850017529)}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 0964d4332c

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".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[jbeckwith-oai]

Independent two-pass review: Vercel ecosystem routes enforce fail-closed bearer authentication, streamed request bodies honor the byte limit regardless of Content-Length, and uploaded files are cleaned up across success/error paths. The earlier request-size review finding is fixed. No actionable findings.