mcphub-operator
Pass
Audited by Gen Agent Trust Hub on Jul 7, 2026
Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes local scripts and system commands for validation and preflight checks.
- Evidence: "scripts/check.py" uses "subprocess.run" to call "doctor.py" and other toolkit scripts.
- Evidence: "scripts/preflight.sh" is a shell wrapper that executes the doctor script using "python3".- [REMOTE_CODE_EXECUTION]: The validation logic includes dynamic loading of repository modules based on the current execution path.
- Evidence: "scripts/doctor.py" modifies "sys.path" and uses dynamic imports to load configuration helpers from the parent repository. This is a standard pattern for local developer tools.- [EXTERNAL_DOWNLOADS]: The documentation and scripts refer to external tools and package managers required for repository management.
- Evidence: The skill refers to "just" for task execution, "uv" for Python environment management, and the upstream "mcphub" CLI tool.
- Evidence: "scripts/check.py" attempts to locate an external asset toolkit in a parent directory.- [DATA_EXFILTRATION]: The skill interacts with local environment variables and configuration files containing sensitive path references.
- Evidence: "scripts/doctor.py" checks for the presence of "MCPHUB_BEARER_TOKEN" in the environment to verify configuration status.
- Evidence: "SKILL.md" and "references/compression-opt-in.md" provide guidance for managing local-only credentials in ".env.mcphub" and "~/.cloudflared/".
- Note: No network exfiltration of data was detected; the skill explicitly reinforces the practice of keeping credentials local and out of version control.
Audit Metadata