generate-agents-md
Pass
Audited by Gen Agent Trust Hub on Feb 20, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [NO_CODE] (SAFE): This skill is composed entirely of Markdown instructions and templates for the AI agent. No executable scripts (Python, Node.js, etc.) or external package dependencies were detected.
- [DATA_EXFILTRATION] (SAFE): No network operations, API calls, or logic for transmitting data to external servers were found.
- [PROMPT_INJECTION] (SAFE): No malicious instructions designed to bypass agent safety filters or override system prompts were detected in the skill content or metadata.
- [COMMAND_EXECUTION] (SAFE): While the documentation mentions bash commands as examples for the generated file, the skill itself does not programmatically execute any commands or scripts.
- [INDIRECT_PROMPT_INJECTION] (SAFE): An analysis of the indirect injection surface was performed. 1. Ingestion points: The skill reads untrusted project files such as CLAUDE.md, README.md, and various package manager files. 2. Boundary markers: None are explicitly defined in the prompt instructions. 3. Capability inventory: The agent's capability is restricted to writing a static Markdown file (AGENTS.md) to the project root; no subprocess, network, or dynamic execution capabilities are used. 4. Sanitization: No explicit sanitization of ingested text is performed. The risk is rated as SAFE because the skill only generates non-executable documentation and does not trigger any high-risk tools.
Audit Metadata