vibe-coding-document
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from user-provided PRDs and existing project codebases to populate documentation templates. A critical risk arises during the generation of the
CLAUDE.mdfile, which serves as a persistent governance and instruction layer for the AI agent. Malicious instructions embedded in source code comments or imported requirements could be promoted into this instruction file, potentially influencing the agent's behavior in future sessions or bypassing safety controls. - Ingestion points: Processes external content through the 'PRD Import Mode' (
references/prd-import.md) and 'Reverse-Engineer Mode' codebase analysis (references/codebase-analysis.md). - Boundary markers: The skill does not implement explicit delimiters or instructions to separate untrusted data from the generated instructions.
- Capability inventory: The skill is capable of reading project files and writing documentation to the local file system (e.g.,
./docs/,CLAUDE.md). - Sanitization: No sanitization or filtering logic is specified for the external content before its interpolation into documentation templates.
Audit Metadata