agentmd-generator
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill analyzes repository files (e.g., config files, READMEs) to generate AI instructions (AGENTS.md), creating a surface where malicious content in the codebase could be promoted to a system-level instruction for future agent sessions. \n
- Ingestion points: Systematic inspection of repository directories and configuration files (package.json, go.mod, pyproject.toml, etc.) as described in Phase 1 and Phase 2 of
SKILL.md. \n - Boundary markers: The templates provided in
assets/root-agents-template.mdandassets/local-agents-template.mdlack delimiters or specific instructions for the agent to ignore potentially malicious instructions embedded within the extracted project data. \n - Capability inventory: The skill requires broad filesystem reading capabilities for analysis and file writing capabilities to generate the AGENTS.md files across the repository structure. \n
- Sanitization: There is no documented mechanism for sanitizing, escaping, or validating the strings extracted from the codebase before they are interpolated into the final instruction templates.
Audit Metadata