mcp-developer
Pass
Audited by Gen Agent Trust Hub on Mar 27, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious patterns, prompt injections, or unauthorized data exfiltration attempts were detected. The skill instructions and reference materials promote secure development practices.
- [PROMPT_INJECTION]: The skill explicitly addresses the indirect prompt injection surface associated with MCP tools and resources. 1. Ingestion points: Arguments in 'tools/call' and content in 'resources/read'. 2. Boundary markers: The protocol uses structured JSON-RPC messages to separate control and data. 3. Capability inventory: Examples include database, filesystem, and API integrations. 4. Sanitization: The skill mandates schema-based validation and provides logic for verifying directory boundaries.
- [COMMAND_EXECUTION]: Reference materials demonstrate tool functions that interact with the system (e.g., Git commands) while providing guidance on implementing path resolution checks to prevent unauthorized access.
- [EXTERNAL_DOWNLOADS]: Dependency references are limited to official, standard SDKs for the Model Context Protocol ecosystem, such as the Python 'mcp' and TypeScript '@modelcontextprotocol/sdk' packages.
Audit Metadata