Agent-Corex – Intelligent Tool Selection
this week
discussions
this week
Agent-CoreX is a production-focused MCP tool retrieval engine and command-line utility for AI agent environments. It targets the common issue of having many available MCP tools/servers while needing to expose only a small, relevant subset to an LLM to avoid excessive context usage. Agent-CoreX runs as an MCP gateway server (stdio) that provides a fixed interface to supported AI clients and performs semantic tool retrieval to select the most relevant tools for a given query. It also routes tool executions to the correct underlying MCP server and performs auth checks against a backend API key. The CLI helps users detect supported AI tools on a machine, inject or remove the Agent-CoreX MCP entry into their configs, browse a registry of installable MCP servers, install/update servers, and run diagnostics to validate Python, PATH, configuration, backend connectivity, and injection status.
LLM agents can’t choose useful tools from large MCP toolsets within context limits
MCP gateway+CLI ranks tools, injects servers, and routes calls via a single entry point
Developers using Claude/Cursor/VS Code with many MCP servers
Share your thoughts about this tool.
Sign in to leave a comment.
No comments yet. Be the first to leave one.