
codemesh
by kiliman·★ 25·Score 41
CodeMesh is a self-improving MCP server that lets AI agents orchestrate multiple MCP servers through TypeScript code.
Overview
CodeMesh provides an innovative approach to MCP server orchestration by exposing just three tools: discover-tools, get-tool-apis, and execute-code. Instead of flooding agents with dozens of individual tools, CodeMesh allows agents to write TypeScript code that calls multiple servers simultaneously. Its key innovation is auto-augmentation, where agents document unclear tool outputs to help future agents succeed in one shot.
Try asking AI
After installing, here are 5 things you can ask your AI assistant:
When to choose this
Choose CodeMesh when you need to orchestrate multiple MCP servers and want the system to self-improve through documentation of tool outputs.
When NOT to choose this
Don't choose CodeMesh if you need fine-grained control over individual tools or if your use case only requires a single MCP server.
Tools this server exposes
3 tools extracted from the READMEdiscover-toolsSee what MCP tools are available from connected servers
get-tool-apisGet TypeScript API definitions for specific tools
execute-codeExecute TypeScript code that calls multiple MCP servers
Comparable tools
Installation
Installation
- Add CodeMesh to Claude Desktop
claude mcp add codemesh npx -y codemeshOr manually add to your Claude Desktop MCP settings:
{
"mcpServers": {
"codemesh": {
"command": "npx",
"args": ["-y", "codemesh"]
}
}
}- Create Configuration
Create a .codemesh/config.json file in your project directory to configure which MCP servers CodeMesh should connect to.
FAQ
- How is CodeMesh different from traditional MCP servers?
- Instead of exposing many individual tools, CodeMesh provides just three tools that let agents orchestrate multiple servers through TypeScript code execution. It also auto-documents unclear outputs to improve performance over time.
- What is auto-augmentation?
- Auto-augmentation is CodeMesh's key innovation where agents automatically document unclear tool outputs. This documentation helps future agents understand the output format, enabling one-shot success instead of trial-and-error.
Compare codemesh with
Last updated · Auto-generated from public README + GitHub signals.