The Agent Skills Directory

Persistence Mechanisms (HIGH): The 'Shell Completions Setup' section provides explicit instructions for the agent to modify user shell profiles (~/.zshrc, ~/.bashrc) to source completion files. In the context of an AI skill, modifying shell startup scripts is a high-risk persistence mechanism.
Indirect Prompt Injection (HIGH):
Ingestion points: The skill is designed to bootstrap and manage project repositories by creating justfile modules based on project content.
Boundary markers: None. The skill uses standard markdown templates without delimiters to separate project data from execution logic.
Capability inventory: The skill utilizes the just command runner, which executes arbitrary shell commands including rm -rf, go build, and chmod.
Sanitization: None. If an agent generates these recipes based on untrusted project data (e.g., malicious file names or branch names in git describe), it could lead to arbitrary command execution.
Command Execution (MEDIUM): The templates include high-impact commands such as chmod +x on bootstrap scripts and rm -rf on variable-defined paths, which present a risk if the agent misinterprets the project root or variable scope.

just