Cognite Industrial MCP
MCP endpoint on Cognite Data Fusion that reuses the signed-in user's permissions for graph, time series, and documents.
Who it's for
Operations or data teams who already keep plant context in Cognite. Most of these products do not speak a field bus themselves.
Caveat / limitations
Documented setup copies a JSON config into Cursor, Claude, or Copilot. The endpoint shape is https://{cluster}.cognitedata.com/mcp, bound to one CDF project at sign-in. Tools match what Atlas AI agents use: query, time series, and documents. Access is the user's existing CDF permissions. This is an industrial knowledge graph, not a PLC driver. CDF is a paid platform.
Protocols
No field protocol or IEC language is claimed. If this tool reaches a PLC, it does so through something else already named in the notes.
Sources
- Setup docshttps://docs.cognite.com/cdf/build/mcp/setup
This page: https://otaihub.com/tools/cognite-industrial-mcp
Same category
- HighByte Intelligence Hub MCPCommercial industrial DataOps hub that publishes pipelines you model as MCP tools, including optional write-back.
- Litmus MCP ServerOpen-source MCP plugin over Litmus Edge: devices, tags, telemetry, containers, and a large SDK fallback.
- Proxus platform MCPGoverned MCP server for the Proxus operations product: device health, graph, telemetry, and explicit writes.
- Seeq AIIndustrial analytics suite with an AI assistant and a newer agent layer for operational decisions.