generate-mcp-server

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill guides the agent to use the speakeasy CLI for authentication (speakeasy auth login), specification validation (speakeasy lint), and code generation (speakeasy run). It also provides instructions for running the generated server using npx (e.g., npx petstore-mcp mcp start).
  • [INDIRECT_PROMPT_INJECTION]: The skill acts on external data sources, specifically OpenAPI specifications and YAML configuration files. This presents a surface where malicious descriptions within a specification could attempt to influence an AI assistant that eventually consumes the generated MCP tools.
  • Ingestion points: Reads project configuration files (.speakeasy/workflow.yaml, .speakeasy/gen.yaml) and API specifications (openapi.yaml).
  • Boundary markers: None explicitly mentioned in the instructions; the external Speakeasy CLI handles the parsing and generation logic.
  • Capability inventory: The instructions enable code generation and subsequent execution of the generated package via npx.
  • Sanitization: The skill provides instructions on using overlays to filter or disable sensitive endpoints, which serves as a manual sanitization step for the user.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 07:45 PM
Security Audit — agent-trust-hub — generate-mcp-server