MCP Clients
openfang
Set it running once, and wake up to a finished competitor report in your inbox.
Needs: Node.js >= 18 installed (download)OpenFang installed and initialized
rowboat
Your assistant can read your email, notes, and meetings, then act on them.
Needs:
Qwen-Agent
Ask a Qwen-powered assistant to read your files, run code, and browse pages.
Needs:
cc-haha
Turns Claude Code from a terminal tool into a proper desktop workspace.
Needs:
MiMo-Code
Reads your code, writes fixes, runs commands, and remembers your project across sessions.
Needs:
ironclaw
Your assistant can remember notes, automate tasks, and chat with you on Telegram or Slack.
Needs:
mcp-use
Turns any AI model you like into an agent that can use outside tools.
Needs: Python 3.11+MCP implementation (like Playwright MCP)LangChain and appropriate model libraries (OpenAI, Anthropic, etc.)
voltagent
A TypeScript framework and console for agents that use MCP tools and remember things.
Needs:
omnigent
No more switching between separate apps to track your different AI coding assistants.
Needs:
Upsonic
When an agent's answer has to be right, a second agent checks its work.
Needs: Python 3.10 or higherAccess to OpenAI or Anthropic API keys (Azure and Bedrock Supported)
CodePilot
Lets you chat with any major AI model from one app on your computer.
Needs:
mission-control
Lets you track, assign, and review your AI coding agents' tasks from one dashboard.
Needs:
LobsterAI
Lets you run AI agent tasks on your desktop and control them from your phone.
Needs:
klavis
Add ready-made tools to your AI agents without hosting anything yourself.
Needs:
holaOS
Halfway through a coding task, switch from Claude to Codex without losing context.
Needs: