ck:mcp-management

MCP server discovery and tool execution can be managed without expanding the main context window by delegating work to subagents or using generated catalogs. This approach keeps the primary conversation compact while handling multi-server workflows automatically.

Multi-server discovery is the process of connecting to several MCP servers from one configuration to inspect their available tools, prompts, and resources. It enables intelligent execution across web search, memory storage, filesystem access, and browser automation tasks.

Configure your MCP servers and ask the system to inspect them, list available capabilities, and execute the matching tool. It handles JSON-RPC client communication, subprocess transport cleanup, and configuration loading to run the right sequence automatically.

Yes, TypeScript is supported for MCP server management alongside the Model Context Protocol SDK. The environment uses ts-node, tsx, and nodemon to handle JSON-RPC client interactions and subprocess transport cleanup for reliable operation.

Yes, MCP tool orchestration supports web search, memory storage, filesystem access, and browser automation. It discovers the available tools across connected servers and intelligently selects and calls the right capability for your specific task sequence.

Subprocess transport cleanup is required to ensure reliable MCP server operation by properly managing JSON-RPC client connections. Without it, abandoned subprocesses and lingering transports would cause unpredictable tool execution failures across multi-server workflows.