The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the fastmcp CLI tool to list, call, and discover tools provided by MCP servers. This involves executing shell commands to interact with local and remote services.
[EXTERNAL_DOWNLOADS]: The skill documents the use of npx to download and run packages from the official @modelcontextprotocol organization on the NPM registry. These packages are well-known resources for implementing MCP servers.
[DYNAMIC_EXECUTION]: The skill supports executing local files, such as server.py, to launch MCP servers. It also allows running arbitrary stdio commands via the --command flag.
[INDIRECT_PROMPT_INJECTION]: The skill provides an interface to ingest data (tool definitions, schemas, and results) from external MCP servers via HTTP or stdio commands.
Ingestion points: fastmcp list and fastmcp call outputs (SKILL.md).
Boundary markers: None explicitly mentioned to separate untrusted tool output from agent instructions.
Capability inventory: Includes network operations (HTTP servers), file execution (local Python files), and command execution (npx).
Sanitization: No explicit sanitization or validation of the data received from external servers is documented in the instructions.

fastmcp-client-cli