The Agent Skills Directory

[Prompt Injection] (SAFE): No attempts to override system safety filters or bypass instructions were detected. The instructional language (e.g., 'NEVER', 'ALWAYS') is used exclusively to reinforce the search methodology.
[Data Exposure & Exfiltration] (SAFE): The skill does not perform network operations or access sensitive system paths (e.g., SSH keys, credentials). It uses local tools like grep and find for project-specific exploration.
[Unverifiable Dependencies & Remote Code Execution] (SAFE): No external packages (npm/pip) or remote scripts are downloaded or executed. Allowed tools are restricted to standard local utilities.
[Obfuscation] (SAFE): No encoded strings, zero-width characters, or homoglyphs were found in the skill definition.
[Indirect Prompt Injection] (LOW): The skill possesses a vulnerability surface common to search tools where it ingests untrusted data from a codebase.
Ingestion points: Processes external files using grep and file: read commands.
Boundary markers: None specified for the search results or file content.
Capability inventory: Limited to grep, find, git, and read tools.
Sanitization: No explicit sanitization or filtering of file content is mentioned before the agent processes it.

zero-in