mcp-server-patterns
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill contains technical instructions for developers and does not attempt to override agent behavior or bypass safety filters.
- [DATA_EXPOSURE_AND_EXFILTRATION]: No sensitive file paths, hardcoded credentials, or unauthorized network operations were identified.
- [OBFUSCATION]: No hidden content, encoded strings, or homoglyphs were detected in the documentation.
- [UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill references the official MCP SDK and Zod validation library, both of which are standard and reputable resources. It does not include commands for remote script execution or downloads from untrusted sources.
- [INDIRECT_PROMPT_INJECTION]: While the skill describes the ingestion of untrusted data through tool and resource handlers, it explicitly recommends using Zod for strict schema validation and structured error handling to mitigate potential injection risks.
Audit Metadata