mcp-server-scaffold

Pass

Audited by Gen Agent Trust Hub on Sep 12, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the creation of a server skeleton based on user-supplied definitions. This ingestion of external data constitutes an attack surface if the input is not sanitized before being used in code generation.
  • Ingestion points: Tool names, input schemas, and error contracts provided by the user in response to steps 1-3 in SKILL.md.
  • Boundary markers: The instructions lack specific delimiters or warnings to ignore malicious instructions embedded within user-provided schemas.
  • Capability inventory: The skill involves generating Python code (using FastMCP) and executing it locally for smoke testing as described in steps 5 and 6 of SKILL.md.
  • Sanitization: No sanitization or validation steps for user-provided strings are included in the scaffolding process.
  • [COMMAND_EXECUTION]: Step 6 of SKILL.md instructs the agent to run the generated server locally for smoke testing. This involves executing code that incorporates user-provided schemas and descriptions.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 12, 2026, 06:21 AM
Security Audit — agent-trust-hub — mcp-server-scaffold