Roboto MCP: robotics data for AI agents
Roboto MCP
The hosted Roboto MCP server lets an agent search a robot fleet's datasets and files and analyze topic data without changing them
- 58 ★ on the repository
- Last commit
- MPL-2.0
Extended access
Why this level
- The agent reads your organization's test data and robot logs in the Roboto cloud service
- Calls run with your account's permissions, so a token with broad access shows the agent every dataset
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add mcp/roboto-mcpDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The agent reads your organization's test data and robot logs in the Roboto cloud service
- Calls run with your account's permissions, so a token with broad access shows the agent every dataset
This is third-party code. Review the repository files before installing.
What it does
Roboto is an analytics platform for robotics and physical AI. It ingests ROS 1, ROS 2, PX4 and ArduPilot logs, plus Parquet, CSV, video and system journal logs. The hosted Roboto MCP server at https://mcp.roboto.ai/mcp connects Claude Code, Claude Desktop, Codex, Cursor, VS Code and any MCP-capable client to that data. The agent searches the data, browses datasets and files and analyzes topic data right in the conversation. Sign-in happens as yourself, through the browser or with a personal access token, and every call runs with your permissions. The toolset is read-only, and nothing exposed over MCP can modify your data. If you belong to several organizations, one connection covers them all: the agent calls whoami and set_active_org to switch. The same repository holds the roboto Python SDK and the documentation of the roboto command-line utility, which ships as separate binaries.
Who it is for. Engineers on robotics teams who keep flight and drive logs in Roboto and want to work through incidents with an agent.
Good fit when
- After a failed drive you need every run in the fleet logs with the same error message
- An agent should browse a week of datasets and files and summarize what changed in the topics
- An analyst works in Cursor and wants to ask questions of Roboto data without opening the web app
- One connection should cover several organizations that the agent switches between itself
- A team wants to give an agent data access that cannot modify the data
Not a fit when
- Your data is not in Roboto, since the server works only with that platform
- The agent must upload files, run actions or change tags, while the MCP tools only read
- You want local bag-file analysis without the cloud, which the SDK and command-line utility fit
- Company policy forbids sending test data to a cloud service
Example request
Find runs in last week's datasets where the distance sensor topic produced zeros and describe what they have in common.MCP
- Transport
- http
- Authentication
- OAuth
Limitations
A Roboto account at app.roboto.ai with data in an organization is required. The server is hosted by Roboto and queries go through the cloud. The README does not list every tool and points to the documentation at docs.roboto.ai. Uploading data and running platform actions goes through the Python SDK, which needs Python 3.10 or newer, or through the web app. Where browser sign-in is not possible, such as in CI, an access token is created in the web app.
How to disable. Run claude mcp remove roboto. A personal access token is revoked in the Roboto web app, and the SDK is removed with pip uninstall roboto.
FAQ
How do I connect Roboto MCP to Claude Code?
Run claude mcp add --transport http roboto https://mcp.roboto.ai/mcp and sign in to Roboto through the browser. Instructions for other clients are collected in the Roboto documentation at docs.roboto.ai.
Can the agent change my data through MCP?
No, the toolset is read-only: search, retrieval and analysis. Calls run with your own permissions, so the agent cannot see what you cannot access.
Which data formats does Roboto understand?
ROS 1 (.bag), ROS 2 (.mcap, .db3), PX4 (.ulg), ArduPilot (.bin, .log, .tlog), Parquet, CSV, video and journal logs. Each format has an ingestion action in the Action Hub.
Do I need the SDK for MCP to work?
No, the hosted MCP server connects with a single command. The SDK installs separately with pip install roboto and is needed for uploading and processing data from Python.
What do I install to upload data from Python?
The roboto package with pip install roboto, then sign in with python -m roboto.cli auth login. Heavier parts come as extras: roboto[analytics] for pandas and similar-signal search, roboto[ingestion] for Parquet, roboto[video] for video frames.
Discussion
No questions or reviews yet.
Related
MCP serversAn MCP server that answers plain-language questions about robot, drone and IoT logs, with each calculation shown as a SQL query
MCP server that connects Claude Code and Codex to an open R or Python session in RStudio and Positron, where your data lives
The official Graphlit MCP server: an agent collects material from Slack, GitHub, mail and websites into a knowledge base and searches it
The Airbyte MCP connects an agent to an Airbyte organization: pipelines, syncs, job logs and direct data reads.
An agent browses FiftyOne datasets, runs aggregations and operators and drives the app UI, with more than 45 tools
An MCP server with a toolbox for global briefs: markets, SEC filings, conflict, climate and news from open sources
README in short
The README presents the Roboto Python SDK and command-line utility, showing browser sign-in and uploading a file to a dataset. The section Connect Your AI Tools covers the hosted MCP server, followed by a table of supported data formats.