The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the gh (GitHub CLI) tool for project management tasks, including gh issue list for searching existing issues, gh label create for organizing feedback by area, and gh issue create for filing findings. These operations are essential for the skill's intended functionality of repo maintenance.
[INDIRECT_PROMPT_INJECTION]: The skill's input is unstructured user feedback, which represents a surface for indirect prompt injection. However, the risk is mitigated by the design of the skill, which requires the agent to verify claims against the actual codebase and documentation (ADRs, git history) and mandates a clarification step with the user. This human-in-the-loop and evidence-based approach prevents the direct execution of malicious instructions embedded in user comments.
Ingestion points: Unstructured user feedback provided via natural language.
Boundary markers: None explicitly defined for the initial user input.
Capability inventory: GitHub CLI operations (gh label create, gh issue create) and repository analysis tools.
Sanitization: The skill instructs the agent to transform user feedback into a "short slug" and specific numbered points, acting as a logical filter between the raw input and the final commands.

feedback