mcp-setup
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONNO_CODE
Full Analysis
- [PROMPT_INJECTION] (LOW): The skill configures MCP servers that ingest data from untrusted external services (e.g., GitHub, Notion, Slack), creating a surface for Indirect Prompt Injection.
- Ingestion points: External APIs, databases, and filesystems.
- Boundary markers: None specified in the setup instructions.
- Capability inventory: MCP tools can perform actions like database writes and API calls.
- Sanitization: No sanitization or validation of external content is mentioned in the configuration steps.
- [EXTERNAL_DOWNLOADS] (LOW): The skill provides examples of downloading and executing packages from the npm registry using
npx -yfor local stdio servers. - [COMMAND_EXECUTION] (LOW): The skill involves executing shell commands (
claude mcp add) to modify local system configuration files (~/.claude.json).
Audit Metadata