nemoclaw-maintainer-cross-issue-sweep
Pass
Audited by Gen Agent Trust Hub on May 18, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill performs repository analysis using the GitHub CLI (gh). It retrieves Pull Request diffs and issue data through authenticated API calls, which is consistent with its stated purpose of managing a codebase's issue queue.\n- [PROMPT_INJECTION]: Indirect injection surface. The skill processes untrusted content from GitHub PR and issue fields (ingestion points: checks/relationship-judgment.md). While it does not utilize explicit boundary delimiters for all external text, it mitigates risk through a strict evidence requirement and deterministic pre-processing. The agent has shell and Python execution capabilities (capability inventory: scripts/ directory) used for structured data processing. Sanitization is performed during fingerprint extraction to filter out generic tokens and language keywords.\n- [COMMAND_EXECUTION]: Local script execution. The skill executes locally provided bash and Python scripts to automate data extraction and report rendering. These scripts include input validation, such as verifying that PR numbers are numeric, to prevent command injection during local execution.
Audit Metadata