SkillsMP API
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
- [EXTERNAL_DOWNLOADS] (HIGH): The skill identifies and encourages the download of external AI skills from arbitrary third-party repositories discovered via the SkillsMP API. This bypasses typical repository vetting processes.
- [REMOTE_CODE_EXECUTION] (HIGH): The core workflow (documented in SKILL.md and README.md) guides the agent to use 'npx add-skill' to install and execute code found during search. This is a high-risk capability when processing untrusted marketplace data. (Evidence: Ingestion at scripts/install_helper.py, Capability at SKILL.md installation workflow).
- [CREDENTIALS_UNSAFE] (MEDIUM): The 'scripts/setup.py' utility accepts the SkillsMP API key as a command-line argument. This practice is unsafe as it often leaves sensitive credentials in the user's shell history (e.g., .bash_history).
- [COMMAND_EXECUTION] (MEDIUM): The skill documentation instructs agents to execute multiple shell commands including package installations (pip) and remote script executions (npx) without providing sanitization or safety checks for the arguments derived from external API responses.
Recommendations
- AI detected serious security threats
Audit Metadata