OmniMCP
A semantic router that provides unified access to multiple MCP servers through a single tool interface, using vector search to discover and execute tools across your entire MCP ecosystem while minimizing context token usage.
What is OmniMCP?
OmniMCP is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to semantic router that provides unified access to multiple mcp servers through a single tool interface, using vector search to discover and execute tools across your entire mcp ecosystem while minimizin...
A semantic router that provides unified access to multiple MCP servers through a single tool interface, using vector search to discover and execute tools across your entire MCP ecosystem while minimizing context token usage.
This server falls under the Developer Tools category on MCPgee, the world's largest MCP server directory with 33,000+ servers.
Features
- A semantic router that provides unified access to multiple M
Use Cases
Maintainer
Works with
Installation
Manual Installation
npx omnimcpConfiguration
Configuration Details
claude_desktop_config.json
Performance
Response Metrics
Resource Usage
How to Set Up and Use OmniMCP
OmniMCP is a Python-based computer-use agent framework from OpenAdaptAI that combines visual UI understanding (via OmniParser) with LLM-driven planning to autonomously interact with any graphical desktop application. It operates through a perceive-plan-act loop: it takes a screenshot, identifies UI elements semantically, generates an action plan using Claude or another LLM, and executes mouse and keyboard inputs via pynput. Unlike purely DOM-based automation, OmniMCP works on any native or web application without browser-specific dependencies, making it suitable for tasks like form filling, multi-step workflows, and cross-application automation.
Prerequisites
- Python 3.10 or later
- An Anthropic API key (ANTHROPIC_API_KEY) for Claude-powered planning
- AWS credentials (AWS_ACCESS_KEY_ID and AWS_SECRET_ACCESS_KEY) if deploying OmniParser on EC2
- An MCP-compatible client such as Claude Desktop
- macOS or Linux desktop environment (pynput requires display access)
Clone the repository
Clone the OmniMCP repository from GitHub to your local machine.
git clone https://github.com/OpenAdaptAI/OmniMCP.git
cd OmniMCPRun the install script
Execute the provided install script, which creates a Python virtual environment and installs all dependencies including pynput and the OmniParser integration.
./install.shCopy and configure the environment file
Copy the example environment file and fill in your API keys. ANTHROPIC_API_KEY is required; AWS credentials are only needed if you want OmniMCP to deploy an OmniParser server on EC2 automatically.
cp .env.example .env
# Edit .env and set:
# ANTHROPIC_API_KEY=your_anthropic_key
# AWS_ACCESS_KEY_ID=your_aws_key (optional)
# AWS_SECRET_ACCESS_KEY=your_aws_secret (optional)
# OMNIPARSER_URL=http://your-parser:8000 (optional custom endpoint)Activate the virtual environment
Activate the Python virtual environment created by the install script before running the agent.
source .venv/bin/activateRun the agent with a goal
Launch the CLI with a natural-language goal. The agent will open the target application, perceive the screen, plan steps, and execute them autonomously. The default demo opens Calculator and computes 5×9.
python cli.py --goal "Open Safari and search for the latest news about Model Context Protocol"Enable debug logging if needed
Set LOG_LEVEL=DEBUG in your .env file to see detailed per-step output including the parsed UI elements, generated action plan, and executed inputs.
LOG_LEVEL=DEBUG python cli.py --goal "Fill out the contact form at example.com"OmniMCP Examples
Client configuration
Register OmniMCP as an MCP server in Claude Desktop by pointing to the Python entry point inside the virtual environment.
{
"mcpServers": {
"omnimcp": {
"command": "/path/to/OmniMCP/.venv/bin/python",
"args": ["cli.py"],
"env": {
"ANTHROPIC_API_KEY": "your_anthropic_api_key",
"OMNIPARSER_URL": "http://localhost:8000"
}
}
}
}Prompts to try
Goals you can pass to OmniMCP via the CLI or through your MCP client.
- "Open the Calculator app and compute 128 divided by 16"
- "Fill in the login form at http://localhost:3000 with username 'admin' and password from my password manager"
- "Open Finder, navigate to ~/Documents, and rename all files containing 'draft' to append '_final'"
- "Open System Settings and enable Dark Mode"Troubleshooting OmniMCP
pynput raises PermissionError or inputs are not delivered
On macOS, grant Accessibility permissions to Terminal (or your IDE) under System Settings > Privacy & Security > Accessibility. On Linux, ensure your user is in the 'input' group or run with appropriate display permissions.
OmniParser fails to start or OMNIPARSER_URL is unreachable
If you are not using the AWS auto-deploy, start the OmniParser service locally by following its own README and set OMNIPARSER_URL=http://localhost:8000 in your .env file.
The agent loops or fails to complete a goal
Set LOG_LEVEL=DEBUG and review the per-step plan output. Simplify the goal into smaller sub-tasks. Verify that ANTHROPIC_API_KEY is valid and that your account has sufficient quota.
Frequently Asked Questions about OmniMCP
What is OmniMCP?
OmniMCP is a Model Context Protocol (MCP) server that semantic router that provides unified access to multiple mcp servers through a single tool interface, using vector search to discover and execute tools across your entire mcp ecosystem while minimizing context token usage. It connects AI assistants to external tools and data sources through a standardized interface.
How do I install OmniMCP?
Follow the installation instructions on the OmniMCP GitHub repository. Clone the repo, install dependencies, and add the server config to your AI client.
Which AI clients work with OmniMCP?
OmniMCP works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.
Is OmniMCP free to use?
Yes, OmniMCP is open source and available under the MIT License license. You can use it freely in both personal and commercial projects.
OmniMCP Alternatives — Similar Developer Tools Servers
Looking for alternatives to OmniMCP? Here are other popular developer tools servers you can use with Claude, Cursor, and VS Code.
Ecc
★ 188.2kThe agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
Javaguide
★ 155.8kJava 面试 & 后端通用面试指南，覆盖计算机基础、数据库、分布式、高并发、系统设计与 AI 应用开发
Gemini CLI
★ 104.5kA secure MCP server that wraps the Google Gemini CLI, allowing clients to query Gemini models using local OAuth sessions without requiring an API key. It provides tools for model interaction and diagnostics with built-in protection against command in
Awesome MCP Servers
★ 87.3k⭐ Curated list of Model Context Protocol (MCP) servers - tools that extend Claude Desktop, Cursor, Windsurf, and other MCP clients with custom capabilities.
MCP Servers
★ 86.0kModel Context Protocol Servers
CC Switch
★ 77.5kA cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Gemini CLI & Hermes Agent. Only official website: ccswitch.io
Browse More Developer Tools MCP Servers
Explore all developer tools servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.
Set Up OmniMCP in Your Editor
Choose your AI client for step-by-step setup instructions.
Quick Config Preview
Add this to your claude_desktop_config.json or .cursor/mcp.json
Ready to use OmniMCP?
Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.