The Agent Skills Directory

[PROMPT_INJECTION]: The skill interacts with external, potentially untrusted content from GitHub issues and local repository files. This presents a surface for indirect prompt injection, but the risk is managed by core safety instructions.
Ingestion points: Reads issue bodies, comments, and repository files via gh issue view and file system access.
Boundary markers: The instructions do not use specific delimiters to isolate external data from the prompt.
Capability inventory: The agent can perform write operations including creating, editing, and closing issues using the gh CLI.
Sanitization: The skill mandates a human-in-the-loop workflow, explicitly requiring the agent to display proposed changes and obtain user approval before executing any mutation commands.
[COMMAND_EXECUTION]: All repository operations are performed using the standard GitHub CLI (gh). The commands used (e.g., gh issue create, gh issue edit) are transparent and strictly limited to the skill's declared purpose of issue management.

issues