The Agent Skills Directory

[SAFE]: The skill demonstrates no malicious behavior, obfuscation, or unauthorized access. It uses standard search tools for its primary research function without escalating privileges or exfiltrating data.
[PROMPT_INJECTION]: The skill contains a surface for indirect prompt injection as it ingests and processes content from external files. 1. Ingestion points: Code files identified via Glob and Grep based on user $ARGUMENTS in SKILL.md. 2. Boundary markers: None present. 3. Capability inventory: Limited to Glob, Grep, and Read tools (read-only filesystem access). 4. Sanitization: No sanitization or content validation is performed on the data read from files.

deep-research