openwiki-update
Pass
Audited by Gen Agent Trust Hub on Jul 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONCREDENTIALS_UNSAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use local tools like
git(status, diff, log, rev-parse, show),rg(ripgrep), andcatfor codebase discovery and metadata management. These commands are typical for documentation tasks and operate within the repository scope. - [PROMPT_INJECTION]: The skill includes persona-setting instructions ('You are acting as an expert technical writer...') and negative constraints ('Do not invent files...', 'Do not search outside the target repository'). These guide the agent's behavior to maintain documentation quality and do not attempt to bypass core safety filters.
- [CREDENTIALS_UNSAFE]: The skill contains explicit 'Security and privacy rules' that prohibit the agent from reading or documenting sensitive information such as secrets, private keys, tokens, or environment files (.env). This provides a defensive layer to prevent accidental credential exposure during the documentation process.
Audit Metadata