agents-md-generator
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests arbitrary data from a target repository (manifests, scripts, configuration files) to generate instructions, which constitutes a surface for indirect prompt injection.
- Ingestion points: Files throughout the repository including manifests (e.g., package.json), READMEs, existing instruction files (e.g., CLAUDE.md), and other documentation or configuration files.
- Boundary markers: The instructions do not define specific delimiters or "ignore" instructions for the data being read, though it advises identifying secret locations without reading their values.
- Capability inventory: The skill uses tool calls to edit files and execute shell commands for validation (e.g., manifest commands, task runners, scripts).
- Sanitization: The skill relies on the user to authorize "consequential validation" commands, but does not specify automated sanitization of the content extracted from the repo.
- [COMMAND_EXECUTION]: The skill is designed to identify and execute shell commands found in the repository (e.g., test runners, build scripts) to verify they are accurate for the AGENTS.md file.
- [EXTERNAL_DOWNLOADS]: The workflow allows for network operations and dependency installation during the validation phase if such actions are defined in the repository being audited.
Audit Metadata