mcp-tool-developer
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions and examples do not contain any malicious patterns such as prompt injection, unauthorized data access, or obfuscation.
- [EXTERNAL_DOWNLOADS]: Mentions official Model Context Protocol SDKs for Node.js (@modelcontextprotocol/sdk) and Python (mcp). These are well-known libraries from trusted sources used for server implementation.
- [CREDENTIALS_UNSAFE]: Explicitly instructs users and the agent to avoid hardcoding API keys or credentials, recommending environment variables and secret managers instead.
- [INDIRECT_PROMPT_INJECTION]: The skill demonstrates awareness of indirect prompt injection vulnerability surfaces. 1. Ingestion points: External API responses and user-provided inputs to tools being developed. 2. Boundary markers: Schema definitions and tool descriptions are recommended to guide the agent. 3. Capability inventory: Guidance covers tools with potential network and file access. 4. Sanitization: Explicitly mandates input validation and treating retrieved text as untrusted.
Audit Metadata