local-clawsweeper-review

Pass

Audited by Gen Agent Trust Hub on Aug 1, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATION
Full Analysis
  • [COMMAND_EXECUTION]: Employs standard developer tools like git, pnpm, and the gh CLI for repository interaction. The skill explicitly limits these to read-only operations unless a user provides a specific request for mutation, adhering to the principle of least privilege.\n- [EXTERNAL_DOWNLOADS]: Retrieves pull request metadata via GitHub APIs and manages project dependencies through pnpm. These actions are restricted to the local development environment and target official, well-known sources.\n- [CREDENTIALS_UNSAFE]: While the skill references authentication tokens (e.g., OpenAI and Codex keys), it enforces secure management practices by instructing the agent to use stdin redirection and environment cleanup rather than persistent storage or insecure logging.\n- [DATA_EXFILTRATION]: Reads locally generated review reports to provide summaries. The instructions include a strict safety boundary that forbids the output of sensitive authentication material or credentials identified during the process.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 1, 2026, 08:33 AM
Security Audit — agent-trust-hub — local-clawsweeper-review