The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted data from the codebase being analyzed and presents it to the agent's context for synthesis.
Ingestion points: The scripts/gather.py script reads source code files from the user-provided repository path via the _source_bytes function. Similarly, scripts/check.py reads _FEATURES.md files from the repository via the parse_features_file function to validate documentation against the code.
Boundary markers: While the output uses markdown headers and code blocks to separate file contents, it does not include explicit instructions to the agent to disregard instructions found within the scanned source code excerpts.
Capability inventory: The skill is designed for agents that possess file-writing capabilities to generate or update documentation files based on the ingested data.
Sanitization: The skill does not perform sanitization or filtering of the ingested content before outputting it to the agent.
[EXTERNAL_DOWNLOADS]: The skill documentation recommends installing the tree-sitter-language-pack package, which is a standard library used for AST-based code scanning.

featuring