The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were detected. The skill is designed for repository maintenance and follows security best practices.
[COMMAND_EXECUTION]: The skill utilizes standard development tools like git and gh (GitHub CLI) to manage branches and create Pull Requests. All mutating actions are explicitly scoped to the .github/ directory and require manual user approval after viewing a diff.
[PROMPT_INJECTION]: While the skill processes repository content (workflow files) that could contain malicious instructions, it mitigates this risk by requiring human review of all changes (git diff) and ensuring that all automated Pull Requests are created as drafts for manual verification.
Ingestion points: Reads files within the .github/ directory (Step 2.3).
Boundary markers: Explicitly requires showing a git diff and getting user confirmation before every commit (Rules, Step 2.4).
Capability inventory: Uses git (checkout, add, commit, push, diff), gh (PR creation), and file system tools (Read, Edit, Glob, Grep) (SKILL.md).
Sanitization: The skill instructions use heredocs and clear templates for the Pull Request body to avoid unexpected command interpolation (Step 3).

action-remediate