mcp-qa
Pass
Audited by Gen Agent Trust Hub on Aug 5, 2026
Risk Level: SAFECOMMAND_EXECUTIONCREDENTIALS_UNSAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute several shell commands using
pnpmto build the project (pnpm -w run build), run lints and tests (pnpm run lint && pnpm run test), and start the MCP server CLI. This is standard behavior for a development and QA utility. - [CREDENTIALS_UNSAFE]: The skill manages authentication tokens by caching them in
~/.sentry/mcp.json. This is a vendor-specific configuration path used for device-code authentication with Sentry's services. - [DATA_EXFILTRATION]: The skill facilitates network communication with production Sentry API endpoints and local development servers (e.g.,
http://localhost:5173) to verify tool functionality and retrieve real production data for testing purposes. - [PROMPT_INJECTION]: The skill contains an indirect prompt injection surface as it is designed to ingest and process arbitrary prompts to exercise MCP tool logic. The instructions encourage using 'realistic prod prompts' to verify the behavior of tools that interact with production data.
Audit Metadata