set-up-project
Pass
Audited by Gen Agent Trust Hub on Oct 20, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a local verification script (
python3 <set-up-machine skill>/scripts/verify.py) to ensure the machine environment is correctly configured. It also utilizes standard CLI tools likegit,gh(GitHub CLI), andglab(GitLab CLI) for repository and issue management. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the repository's issue tracker to assist in documentation setup.
- Ingestion points: Reads existing issue titles from GitHub or GitLab via CLI tools in Step 3 to seed semantic prefixes.
- Boundary markers: The data is formatted into markdown lists within documentation files (
docs/agents/issue-tracker.md). - Capability inventory: The skill has the capability to write files, execute git commands, and perform CLI-based issue tracker operations.
- Sanitization: The skill relies on standard markdown formatting and requires explicit user confirmation for all proposed file modifications, providing a human-in-the-loop safety check.
- [SAFE]: The skill includes a sophisticated security audit mechanism described in
references/project-files.md. This feature scans project-specific harness configurations (e.g.,.claude/settings.json,.cursor/cli.json) to detect and report entries that might weaken the machine's global security rails or safety boundaries.
Audit Metadata