nemoclaw-maintainer-cross-issue-sweep

Pass

Audited by Gen Agent Trust Hub on Jun 18, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill uses the GitHub CLI (gh) to interact with repository data. Accessing a well-known service like GitHub for its intended purpose is considered safe.
  • [COMMAND_EXECUTION]: The skill executes internal shell and Python scripts provided within the skill package (e.g., scripts/extract-fingerprint.sh, scripts/search-candidate-issues.sh). These scripts perform deterministic data extraction and do not execute arbitrary or obfuscated code.
  • [DATA_EXFILTRATION]: No sensitive local data, such as credentials, environment variables, or SSH keys, is accessed or transmitted. The skill's network activity is limited to standard GitHub API interactions via the CLI.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from GitHub issues and PR descriptions (ingestion points: gh pr view, gh issue view). While this presents an indirect prompt injection surface, the risk is low as the skill's capabilities are restricted to analysis and reporting, and the prompt construction in checks/relationship-judgment.md utilizes clear boundary markers for untrusted data.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 18, 2026, 05:27 AM
Security Audit — agent-trust-hub — nemoclaw-maintainer-cross-issue-sweep