The Timebase Historian has a built-in HTTP MCP Server for use with AI Agents and Assistants
What is MCP?
MCP, or Model Context Protocol, is an open standard developed by Anthropic that allows AI models, like Gemini or GPT4, to interact with external tools and data sources, like Timebase Historian. AI models work best when they have "context" - think of the tags, their metadata, and their data as being the "context", MCP as the connector or "protocol", and Gemini as the AI "model". The embedded MCP Server in Timebase Historian exposes "tools" to AI Agents and Assistants to access and analyze time-series data directly.
Where to find it?
The MCP Server is built directly into the Timebase Historian and exposes a number of "tools" and "prompts" for compatible AI Agents and Assistants to make use of. Timebase Historian implements a HTTP MPC Server.
The "entry point" to the MCP is the following address:
URL http://<yourHistorianAddress>:4511/mcp
Here is an example connection using Postman:
- In Postman, click "New", then select "MCP"
- Timebase supports Streamable HTTP MCP - make sure to select "HTTP", not "STDIO"
- Enter the URL as specified above
- Click "Connect"
- If your Historian is running and the connection is successful, Postman will access the MCP Server and request information about the "tools" and "prompts" that are exposed:
How to use it?
More and more AI Agents and Assistants are adding support for MCP Clients every day.
Please refer to the documentation for these MCP Clients to configure and troubleshoot where necessary.
We expect a number of AI Chat applications like ChatGPT will add support very soon. Here are some examples:
Agents
- Claude Desktop by Anthropic
- VS Code by Microsoft
- Cursor
- Windsurf
Assistants
- Copilot Studio by Microsoft
- LibreChat
Frameworks
- OpenAI Agent SDK
- LangGraph
- MCP Bridge
Note: The MCP Server is a Historian feature add-on available to use for free while in BETA. See Licensing.