ROS MCP Server
# ROS MCP Server š§ āš¤
<!-- mcp-name: io.github.robotmcp/ros-mcp-server -->
<p align="center">
<img src="https://github.com/robotmcp/ros-mcp-server/blob/main/docs/images/framework.png"/>
</p>
ROS-MCP-Server connects large language models (such as Claude, GPT, and Gemini) to robots, enabling bidirectional communication with no changes to existing robot source code.
### Why ROS-MCP?
- **No robot source code changes** ā just add the `rosbridge` node to your existing ROS setup.
- **True two-way communication** ā LLMs can both *control* robots and *observe* everything happening on the Robot.
- **Full context** ā publish & subscribe to topics, call services & actions, set parameters, read sensor data, and monitor robot state in real time.
- **Deep ROS understanding** ā guides the LLM to discover available topics, services, actions, and their types (including custom ones) ā enabling it to use them with the right syntax without manual configuration.
- **Works with any MCP client** ā built on the open [MCP standard](https://modelcontextprotocol.io/), supporting Claude Code, Codex CLI, Gemini CLI, Claude Desktop, ChatGPT, Cursor, and more.
- **Works across ROS versions** ā compatible across ROS 2 (Jazzy, Humble, and others) and ROS 1 distros.
---
## š„ Examples in Action
<p align="center">
<a href="https://youtu.be/Yy1loJAn9sA">
<img src="https://github.com/robotmcp/ros-mcp-server/blob/main/docs/images/MCP%20Demos%20Slide%20-%207to12s.gif" alt="ROS MCP demos" />
</a>
</p>
---
š **Example - AI Agent diagnosis of Industrial Robot End Effector** ([Video](https://youtu.be/EhZNFULz9P4))
- The MCP server connects Claude to a production industrial robot, with only the technician manuals as reference.
- Claude discovers the robot's custom topic and service types and their syntax on its own.
- From a single prompt to test the gripper, it reads the manuals, runs its own tests, finds an anomaly, and reports the root cause.
<p align="center">
<a href="https://youtu.be/EhZNFULz9P4">
<img src="https://github.com/robotmcp/ros-mcp-server/blob/main/docs/images/ROS%20MCP%20Gripper%20vacuum%20test.jpg" width="400" alt="Testing and debugging an industrial robot" />
</a>
</p>
---
š¤ **Example - Controlling "Wilson" with natural language** ([video](https://www.traceglarue.com/wilson))
From a single prompt ā *"Grab a Coke from the fridge & go to the living room."* ā Google Gemini uses the MCP server to navigate and manipulate autonomously. Built on ROS 2 with Nav2 (SLAM) for mapping and navigation, and MoveIt to command the manipulator.
<p align="center">
<a href="https://www.traceglarue.com/wilson">
<img src="https://github.com/robotmcp/ros-mcp-server/blob/main/docs/images/Wilson%20thumbnail.jpg" width="400" alt="Wilson robot controlled with natural language" />
</a>
</p>
---
š **Example - Controlling Unitree Go2 in NVIDIA Isaac Sim** ([video](https://www.youtube.com/watch?v=9StFx4lnvmc))
The MCP server connects Claude to a simulated Unitree Go2 quadruped in NVIDIA Isaac Sim, interpreting natural language commands to navigate and control the robot.
<p align="center">
<a href="https://www.youtube.com/watch?v=9StFx4lnvmc">
<img src="https://img.youtube.com/vi/9StFx4lnvmc/maxresdefault.jpg" width="400" alt="Controlling Unitree Go2 in NVIDIA Isaac Sim" />
</a>
</p>
---
## š Getting Started
Follow the [installation guide](docs/install/installation.md) to get started.
ROS-MCP works with Claude Code, Codex CLI, Gemini CLI, Claude Desktop, ChatGPT, Cursor, or any MCP-compatible client.
<p align="center">
<img src="https://github.com/robotmcp/ros-mcp-server/blob/main/docs/images/MCP_topology.png"/>
</p>
---
## š More Examples & Tutorials
Browse our [examples](examples) to see the server in action.
We welcome community PRs with new examples and integrations!
---
## š¤ Contributing
We love contributions of all kinds:
- Bug fixes and documentation updates
- New features (e.g., Action support, permissions)
- Additional examples and tutorials
Check out the [contributing guidelines](docs/contributing.md) and see issues tagged **good first issue** to get started.
---
## š License
This project is licensed under the [Apache License 2.0](LICENSE).
TDQS
Scored across 17 tools
Most tools have distinct purposes focused on different ROS operations like service calls, topic management, and network configuration. However, there is some overlap between get_services and inspect_all_services, as well as between subscribe_once and subscribe_for_duration, which could cause minor confusion in tool selection.
The naming follows a consistent snake_case pattern with clear verb_noun structures like get_topics, call_service, and publish_once. Minor deviations exist, such as ping_robot (which is more specific) and set_websocket_ip (which uses 'set' instead of a more common verb like 'configure'), but overall the conventions are readable and predictable.
With 17 tools, the count is slightly high but reasonable for a ROS server that needs to cover services, topics, messages, and network operations. It provides comprehensive functionality without being overwhelmingly large, though it borders on the upper limit of a well-scoped set.
The tool set offers complete coverage for ROS operations, including CRUD-like actions for services and topics (e.g., call, get, publish, subscribe), message and service details retrieval, and network configuration tools. There are no obvious gaps, and agents can handle core workflows like monitoring, publishing, and service interaction effectively.