openai/openai-agents-python
Overview
This skill runs the mandatory verification stack when changes affect runtime code, tests, or build/test behavior in the repository. It ensures formatting, linting, type checking, and tests all pass before work is marked complete. Use the provided cross-platform scripts or the manual Makefile sequence from the repository root.
How this skill works
The skill executes a fail-fast sequence of commands that enforce code quality and correctness: format, lint, static type checks, and unit tests. It provides cross-platform entry points: a shell script for macOS/Linux and a PowerShell script for Windows, both invoking the required steps in order. If any step fails, stop, fix the issue, and re-run the full stack until all checks succeed.
When to use it
- When changes touch runtime code or business logic
- When tests are added, changed, or failing
- When build or test configuration files are modified
- Before marking a change as complete or merging to main
- Skip only for docs-only or metadata-only changes unless full verification is requested
Best practices
- Run from the repository root to ensure Make targets resolve correctly
- Use the provided scripts (scripts/run.sh or scripts/run.ps1) to enforce correct ordering and fail-fast behavior
- If dependencies changed, run make sync first to install developer requirements
- Do not ignore failures: fix the root cause, then re-run the full stack in order
- Report and include failing output when asking for help to speed diagnosis
Example use cases
- Developer changes a core agent runtime file — run the full stack to confirm no regressions
- A contributor updates unit tests — verify formatting, lint, mypy, and tests pass before merge
- CI pipeline mirrors this skill to gate pull requests that touch code or tests
- A maintainer modifies test configuration — run verification to ensure CI will succeed
FAQ
Only skip for docs-only or repository metadata changes. For any change touching runtime, tests, or build/test config, run the full verification stack.
What if a command fails on my machine but passes in CI?
First run make sync to align dependencies, then re-run the scripts from repo root. If discrepancies persist, capture and share the failing output for investigation.
5 skills
This skill runs the full code-change verification stack to ensure formatting, linting, typing, and tests pass after code changes.
This skill analyzes test coverage results, identifies gaps, proposes high-impact tests, and coordinates approval before implementation.
This skill generates a PR title and draft description after substantive code changes, streamlining PR readiness and collaboration.
This skill audits docs coverage against main branch features and configuration, suggests updates, and reports changes before approving edits.
This skill helps you access authoritative OpenAI API documentation via the MCP server, ensuring accurate guidance from the latest developer docs.
