Labsco
openai logo

code-change-verification

✓ Official★ 3,334

by openai · part of openai/openai-agents-js

Run the mandatory verification stack when changes affect runtime code, tests, or build/test behavior in the OpenAI Agents JS monorepo.

🔥🔥✓ VerifiedFreeQuick setup
#git
🔒 Repo-maintenance skill. It exists to help maintain openai/openai-agents-js itself — it's only useful if you contribute code to that project.

This is the playbook your agent receives when the skill activates — you don't need to read it to use the skill, but it's here to audit before installing.

Code Change Verification

Overview

Ensure work is only marked complete after installing dependencies, building, linting, type checking (including generated declarations), and tests pass. Use this skill when changes affect runtime code, tests, or build/test configuration. This is a post-review final gate: when $implementation-final-review applies, do not invoke the broad stack until its clean-review condition applies to the stable task diff.

Start condition and host capacity

  • During iterative review, use only focused tests and a narrowly targeted static check when the changed typing boundary requires one. Defer repository-wide pnpm -r build-check and the rest of this complete stack until review is clean.
  • Immediately before starting the complete stack, use available read-only task or process evidence to check whether another repository-wide test, typecheck, build, examples runner, or integration command is already active on the same host.
  • When concrete contention is visible, continue useful non-heavy work such as review, remediation, evidence preparation, or focused checks, then check again later. Do not create or wait on a repository lock, host-wide mutex, or sentinel file.
  • Start automatically once review is clean, the diff is stable, and observable host capacity is available. Do not require a user-triggered finalize message. If host telemetry is unavailable, do not block solely because capacity cannot be measured.

Codex execution policy

Repository verification and all child processes must remain in the normal Codex workspace sandbox. Never request elevated sandbox permissions for verification, and never retry with broader host access after a failure.

On macOS/Linux, use the exact Codex command from Quick start so an inherited OPENAI_API_KEY is removed before the repository-controlled wrapper starts. Investigate failures inside the workspace sandbox and report any coverage that the sandbox cannot provide instead of weakening the sandbox boundary.

Manual workflow

  • Run from the repository root in these phases: pnpm i --frozen-lockfile, pnpm build, then pnpm -r build-check, pnpm -r -F "@openai/*" dist:check, pnpm lint, pnpm test, and pnpm format:check:changed.
  • The skill may execute the final validation phase in parallel, but every step above must still pass.
  • Do not skip steps; stop and fix issues immediately when any step fails.
  • Re-run the full stack after applying fixes so the commands execute with the same barriers and coverage.
  • The install intentionally runs without forcing CI or prompt-confirmation settings. If pnpm reports incompatible node_modules, stop and diagnose the configuration mismatch instead of silently recreating dependencies.

Resources

scripts/run.sh

  • Executes the full verification sequence (including declaration checks) with fail-fast semantics.
  • Keeps pnpm i --frozen-lockfile and pnpm build as barriers, then delegates independent validation steps to concurrently.
  • Prefer this entry point to ensure the commands always run from the repo root with the expected fail-fast behavior.

scripts/run.ps1

  • Windows-friendly wrapper that runs the same verification sequence with fail-fast semantics.
  • Use from PowerShell with execution policy bypass if required by your environment.