The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to fetch the latest protocol specifications and SDK documentation from official sources, specifically modelcontextprotocol.io and the associated GitHub repositories. These are well-known resources for the Model Context Protocol.
[COMMAND_EXECUTION]: The evaluation harness provided in scripts/evaluation.py and scripts/connections.py allows the user to execute local shell commands to launch and test MCP servers. This is an intended and documented feature for developers to test their implementations.
[INDIRECT_PROMPT_INJECTION]: The evaluation script ingests questions from user-provided XML files and interpolates them into a prompt template for the LLM. While this is an attack surface for prompt injection, it is a standard design for an evaluation utility, and the script does not perform any autonomous actions without user invocation.

mcp-builder