Isometric provides access to a remote MCP server, allowing suppliers, partners and verifiers to add key context on CDR protocols, Isometric standards and the Certify and Registry platform into common AI chat or coding assistant platforms with support for MCP integration.

​
Available tools

Isometric’s MCP server enables AI assistants to access the following resources in real time to provide context to their responses:
  1. The Isometric Standard
  2. Isometric’s CDR protocols
  3. Isometric’s protocol modules
  4. User guides for the Certify and Registry platform documented on this site
  5. Isometric’s API documentation

​
Connecting to the server

Access to the MCP server is provided using an organization-specific URL like https://api.isometric.com/mcp/?token=***. You will need to be a user of the Certify or Registry platform to generate the URL specific to your organization. You can retrieve it from the Team settings page.
Suppliers can sign up directly for a Certify account here.For MRV partners, research institutions, marketplaces, intermediaries or other organizations, get in touch to request a non-supplier organization account.

​
Supported clients

Remote MCP server integration is supported by most modern AI assistants, including those listed below. Setup instructions are available from the link on each provider. The above list is not exhaustive, any MCP client supporting the Streamable HTTP transport will work with the Isometric MCP server.

​
Setup example (Claude.ai)

1

Manage connectors

On the main chat window, click the Search and tools button and select Manage connectors
2

Add custom connector

In the Connectors settings page, click Add custom connector.
3

Configure custom connector

In the configuration dialog, enter the name of the connector (e.g. “Isometric”) and paste the MCP server URL. Then click the Add button.
Claude Desktop showing the Isometric MCP connectorClaude Desktop showing the response to a question about API usage

Example responses from Claude Desktop with the Isometric MCP connector

​
Suggested prompts

Some suggested prompts to test the MCP server connection are shown below: 
What Isometric-specific capabilities do you have?

How does attribution of emissions work in Certify?

What modules does the Biochar v1.1 protocol reference?

What are the key differences between Biochar v1.0 and v1.1 protocols?

​
Usage tips

​
General advice

If you are unfamiliar with AI assistants, reading educational guides on working with these tools, such as Anthropic’s free AI Fluency course are a good starting point. Assistants can also help guide you on how to get the most out of them: asking them how to work with them effectively as an initial prompt can be a helpful place to begin. If asking multiple different questions, it’s generally best to start a new chat for each question rather than continuing a previous chat, as models usually perform better with shorter conversation histories. If you want to clarify a prompt to get a better answer, editing your original prompt is usually more effective than adding further context in a second prompt.

​
Base prompts

Using a base prompt can help to ensure more reliable behavior from an AI assistant. For chats using the Isometric MCP server, consider using the following base prompt or a similar version:

​
Addressing errors

AI assistants can make errors, particularly if the conversation has a limited context window. If the response from an assistant seems weak, asking for additional evidence can help it to realize where it made a mistake.

​
FAQs

Is my project data shared with the AI provider by connecting the MCP server? In the current implementation of the MCP server, the only additional contextual data is read-only public data on Isometric’s protocol pages and public docs site. Can Isometric read my prompts? Requests are passed to Isometric’s MCP server via a tool call initiated by the AI assistant. The assistant will determine any elements of the prompt that require reference to contexts provided by the MCP server and try to infer what information is required and the intent of the user. An example tool call might look like: 
{
  "name": "Isometric:isometric_docs_get",
  "parameters": {
    "path": "user-guides/certify/create-project"
  }
}
Isometric logs tool calls, which includes which data should be returned to the AI assistant, but the original user prompt is not included.