agents-brain
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill fetches prompting guidance from official documentation at developers.openai.com and platform.claude.com. These operations are managed by a shell script that verifies the target URLs and validates the content markers of the retrieved files.
- [COMMAND_EXECUTION]: Uses a local shell script (scripts/fetch-guidance.sh) and system utilities like ln and git to manage repository context. The script implements defensive measures such as umask 077, atomic file moves, and cache locking.
- [SAFE]: A repository guard rail is implemented in the main instruction file to prevent the skill from operating on the filesystem root or the user's home directory.
- [SAFE]: File modifications are strictly limited to Markdown documentation and symlink creation within a verified Git repository structure.
Audit Metadata