The Agent Skills Directory

[SAFE]: The skill facilitates multi-lens repository auditing by orchestrating existing audit skills. It uses standard git commands to determine repository state and scope without performing dangerous operations.
[INDIRECT_PROMPT_INJECTION]: The skill audits external codebases, which is a surface for indirect prompt injection. Ingestion points: The skill reads all repository files via git ls-files and sub-agent instructions (SKILL.md). Boundary markers: Lacks explicit delimiters for ingested code content but uses structured context blocks for sub-agents. Capability inventory: Spawns concurrent sub-agents and writes report files to the filesystem. Sanitization: No explicit sanitization of codebase content is performed. This risk is inherent to the primary purpose of auditing arbitrary code and is mitigated by the instruction for sub-agents to operate in independent contexts.
[DATA_EXFILTRATION]: The skill includes explicit instructions to verify that the local audit output directory (docs/local/) is listed in .gitignore before writing sensitive findings from the security audit, preventing accidental exposure of vulnerabilities or hardcoded secrets in version control history.

full-review