The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documentation encourages the execution of an external command-line interface tool (faf) to automate project analysis and context synchronization.
Evidence: The 'Fastest path to 100%' section in SKILL.md lists shell commands: faf auto, faf score, faf go, and faf sync.
[INDIRECT_PROMPT_INJECTION]: The skill operates by ingesting untrusted data from the local project environment to generate instructions for AI agents, creating an attack surface where malicious content in project files could influence agent behavior.
Ingestion points: Analyzes the project stack and content from the README file as specified in SKILL.md.
Boundary markers: No specific delimiters or 'ignore' instructions are mentioned for the content being synced into agent configuration files.
Capability inventory: The skill utilizes the faf sync command to write generated context into sensitive agent control files like CLAUDE.md and AGENTS.md.
Sanitization: The documentation indicates the tool seeds facts directly from project strings without describing escaping or validation mechanisms to prevent instruction injection.

faf-context