mcp-builder
Pass
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The guide instructs fetching protocol and SDK documentation from official sources, including modelcontextprotocol.io and github.com. These are established and trusted repositories for MCP development.
- [COMMAND_EXECUTION]: The provided Python scripts (scripts/evaluation.py and scripts/connections.py) allow for the execution of shell commands provided by the user to launch and test MCP servers locally. This is a standard developer tool pattern for software testing.
- [INDIRECT_PROMPT_INJECTION]: The evaluation harness script acts as an agent that incorporates external tool results directly into the message history. This creates an attack surface where a compromised or malicious MCP server could provide instructions to the agent.
- Ingestion points: scripts/evaluation.py processes tool outputs (tool_result) from the connected server.
- Boundary markers: No explicit delimiters or instructions are used to separate tool output from system instructions in the prompt.
- Capability inventory: The harness calls tools on the server and can write reports to the local file system.
- Sanitization: No sanitization or filtering is performed on tool outputs before they are processed by the LLM.
Audit Metadata