isaac-levine/forage

Overview

A TypeScript project that enables AI agents to automatically discover, install, and learn to use new MCP tools. It aims to make tool discovery self-improving so agents can adapt their capabilities without manual intervention.

Best for

Best for
Developers building MCP-based AI agents that need dynamic tool discovery and integration

Use cases

• Enabling an agent to find and add relevant MCP tools on the fly
• Automating the installation of new tools during agent runtime
• Teaching agents how to use discovered tools without prior configuration

How to use

Install

npx forage-mcp init --client cursor

Tools exposed

• forage_search
• forage_evaluate
• forage_install
• forage_learn
• forage_status
• forage_uninstall

Tested with

Claude Code, Cursor

Notes

A TypeScript project that enables AI agents to automatically discover, install, and learn to use new MCP tools. It aims to make tool discovery self-improving so agents can adapt their capabilities without manual intervention.

10 stars on GitHub. Last updated 2026-02-21. Licensed MIT.

Use cases

• Enabling an agent to find and add relevant MCP tools on the fly
• Automating the installation of new tools during agent runtime
• Teaching agents how to use discovered tools without prior configuration

Pros

• Reduces manual setup for tool integration in agent workflows
• Self-improving mechanism allows agents to expand capabilities autonomously
• Open source and focused on MCP ecosystem

Cons

• Very small community (10 stars) suggests limited adoption and support
• Dependent on MCP standard adoption and tool availability
• Requires careful management of agent permissions for automatic installations

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.