The Agent Skills Directory

[SAFE]: The skill operates as a static analysis tool, reading project files and generating a report folder. It does not execute the audited code or perform unauthorized network operations.
[SAFE]: The instructions in SKILL.md and references/security.md specifically direct the agent to identify sensitive credentials (e.g., API keys, JWT secrets) without outputting their actual values, mitigating the risk of accidental data exposure during the audit session.
[PROMPT_INJECTION]: The skill ingests untrusted content from the application source code being audited, creating a surface for indirect prompt injection. 1. Ingestion points: Application source files, configuration files, and environment variable files read during the Recon and Audit phases (SKILL.md). 2. Boundary markers: No explicit delimiters or instructions to ignore embedded commands are used for the audited content. 3. Capability inventory: File-reading to perform the audit and file-writing to create the report folder; the skill lacks network exfiltration or subprocess execution capabilities. 4. Sanitization: No sanitization is performed on the ingested content before processing.

vibe-audit