The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands using nix develop -c mdformat for file formatting and several git commands (status, diff, add, commit) for repository management. These are standard operations for a development workflow.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it analyzes content from markdown files which could contain malicious instructions intended to influence the agent's behavior during the review process. Ingestion points: Markdown files in skills/, .claude/, agents/, templates/, and the current directory. Boundary markers: Absent; content is processed directly. Capability inventory: Shell execution via nix and git. Sanitization: No sanitization of the reviewed content is performed.

commit