The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from local repository configuration files to determine agent affinities for its recommendation engine.
Ingestion points: The init and refresh workflows in reference/cache-recipes.md read from project manifest files including package.json, Cargo.toml, pyproject.toml, and CLAUDE.md.
Boundary markers: No specific delimiters or instructions to ignore potentially embedded prompts are applied during the scanning of these files.
Capability inventory: The skill is restricted to navigation, guidance, and metadata curation; it does not possess network tools or arbitrary code execution capabilities.
Sanitization: No content validation or sanitization is performed on the ingested manifest data before it is used to score skill affinity.
[COMMAND_EXECUTION]: The skill utilizes standard shell commands to analyze the repository structure and size.
Evidence: The init recipe described in reference/cache-recipes.md utilizes find and wc -l to count files for repository sizing and cache optimization logic.

compass