The Agent Skills Directory

[SAFE]: The skill is designed with a strong security posture, explicitly advocating for SHA pinning (to mitigate mutable tag risks) and minimal GITHUB_TOKEN permissions (contents: read) to prevent privilege escalation within CI environments.
[EXTERNAL_DOWNLOADS]: The skill utilizes well-known and reputable external resources, including the 'astral-sh/setup-uv' GitHub Action and the 'zizmor' security linter. These tools are recognized standards in the Python and GitHub Actions ecosystems and are sourced from official repositories.
[COMMAND_EXECUTION]: The skill performs necessary administrative tasks using standard CLI tools such as 'gh api' for repository configuration and 'uvx' for running security linters. The included 'scripts/check_workflows.py' is a benign, read-only script that uses regular expressions to verify the configuration of workflow files without performing any network or destructive operations.
[INDIRECT_PROMPT_INJECTION]: While the skill reads repository metadata (such as 'pyproject.toml') to inform workflow generation—a common pattern that introduces a theoretical injection surface—it provides rigid templates and automated verification scripts that significantly reduce the risk of processing malicious project data.

python-ci