The Agent Skills Directory

COMMAND_EXECUTION (LOW): The skill uses shell commands like mkdir with subshell execution ($(...)) to create dated directories. This is a functional requirement for scaffolding but represents a capability for shell interaction.- EXTERNAL_DOWNLOADS (LOW): The skill recommends cloning external repositories for API review and using npx to run the serve package. These operations involve downloading and interacting with external content.- PROMPT_INJECTION (LOW): Category 8 (Indirect Prompt Injection) risk is present due to the ingestion of untrusted external source code. 1. Ingestion points: External repositories are cloned to scratch/repos/ for analysis. 2. Boundary markers: Absent; there are no instructions provided to the agent to treat external code comments or documentation as untrusted data. 3. Capability inventory: The skill allows for file writing, Python execution via uv, and starting local web servers via npx. 4. Sanitization: No sanitization or validation is applied to the content of cloned repositories before processing.

investigation