The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data by reading source code and manifests from a project directory to perform architectural analysis.
Ingestion points: The skill uses Glob and Grep commands in architecture-guide.md (Step 0) to read file paths and import statements from the codebase.
Boundary markers: The instructions do not define specific delimiters or "ignore embedded instructions" warnings for the data ingested from the project files.
Capability inventory: The skill utilizes codebase exploration tools (Glob, Grep) to populate the agent's context for report generation. No destructive capabilities such as file writing, process spawning, or network access were identified within the skill's logic.
Sanitization: No explicit sanitization or filtering of the content retrieved via Grep or Glob is implemented before the content is processed by the agent.

brooks-audit