The Agent Skills Directory

COMMAND_EXECUTION (MEDIUM): The file scripts/connections.py implements an MCPConnectionStdio class that uses the stdio_client. This client is designed to spawn local subprocesses using a user-provided command and args. While necessary for the skill's purpose of testing local MCP servers, this provides a mechanism for arbitrary command execution.
EXTERNAL_DOWNLOADS (LOW): The SKILL.md instructions guide the agent to fetch documentation from several remote sources, including modelcontextprotocol.io and raw.githubusercontent.com/modelcontextprotocol/. While these appear to be the official protocol sources, they are not on the explicit trusted source list and represent external data ingestion.
PROMPT_INJECTION (LOW): The skill is vulnerable to Indirect Prompt Injection (Category 8).
Ingestion points: The skill explicitly instructs the agent to use WebFetch to load remote markdown files from GitHub and the official MCP website to "study framework documentation."
Boundary markers: There are no instructions for the agent to use delimiters or ignore instructions embedded within the downloaded documentation.
Capability inventory: The skill includes scripts capable of spawning subprocesses (scripts/connections.py) and instructs the agent to run build commands (npm run build) and syntax checks (python -m py_compile).
Sanitization: No sanitization or validation of the remote content is performed before it is processed by the agent.

mcp-builder