KALANOS / MCP

Robotics data insights, in your AI assistant

Search robotics datasets and inspect recorded quality reports with Kalanos MCP.

https://mcp.kalanos.ai/mcp

Ask your data better questions

Discover datasets, compare assessments, and investigate quality issues through your AI assistant.

Answers use available stored reports and do not start new grading jobs. Replace [dataset ID] with your dataset identifier.

Explore datasets

Find robotics datasets for pick-and-place. Show the robot type, episode count, and available quality assessments.

Understand a score

Show the quality report for lerobot/pusht. Explain its score, what was checked, and what wasn’t evaluated.

Compare datasets

Compare lerobot/pusht and lerobot/aloha_sim_insertion_human. Explain differences in quality and evaluation coverage, and whether their scores are comparable.

Investigate problems

For [dataset ID], summarize the most important findings and show the affected episodes where available.

Prioritize review

Which episodes in [dataset ID] should I inspect first? Explain why, using the report’s evidence.

Check suitability

I’m considering [dataset ID] for imitation learning. What does its assessment tell me, and what still needs validation before training?

Dataset missing a report? Submit it for assessment.

Discover datasets

Find public robotics datasets.

Assess training readiness

Check recorded coverage and limitations.

Trace findings to episodes

Follow findings to episode evidence.

Connect with Claude

Open Claude → Customize → Connectors.

+ Add → Add custom connector.

Name it Kalanos. Enter https://mcp.kalanos.ai/mcp, Continue. Review detected authentication, then complete Add.

In a chat, choose + → Connectors and enable Kalanos.

Team / Enterprise workspaces may need owner setup.

Official Claude connector guide

Connect with ChatGPT (web)

Open Plugins → + → Add custom MCP server.

Name it Kalanos. Under Connection, enter https://mcp.kalanos.ai/mcp. Choose No authentication for the current public endpoint.

Review the displayed warning; continue only if you trust the server. Select Create as a plugin.

Review discovered tools and install it. In a new chat, type @, choose Kalanos, and try the example below.

Account / workspace permissions and security restrictions apply. If the option is absent, ask your workspace admin.

Official ChatGPT setup guide

OpenAI custom MCP server guide

Connect with another MCP client

Use a client that supports remote Streamable HTTP. Add a remote MCP server named Kalanos.

Set URL to https://mcp.kalanos.ai/mcp and transport to Streamable HTTP (often called HTTP).

Set authentication to None for public access. No API key is needed.

Save / connect, inspect available tools, enable the server, and run the example below.

Labels and configuration syntax vary by client. Follow your client’s documentation; there is no universal configuration format.

MCP remote-server connection guide

Start with the evidence.

Show the report for lerobot/pusht and explain its scope and limitations.

Scope & available tools

Source: public Hugging Face Space

Browse reports

SOURCE REVIEWED · NOV 3, 2026