MCP · Live
Ask your AI assistant about any company's carbon footprint
The Mycelium MCP server is live. Connect Claude, ChatGPT, Cursor or your own agents to our open-access company emissions database in a couple of minutes. No API key, no account, no sign-up.
01 | What is MCP
An open standard for AI tools and data
Model Context Protocol (MCP) is an open standard that lets large language models talk to external tools and datasets in a structured, secure way.
The Mycelium MCP server exposes our open-access company emissions database as a first-class data source for AI assistants. Ask your model about a company's carbon footprint and it answers with real, sourced figures instead of best guesses, every number tagged as reported by the company or estimated by Mycelium.
We're listed in the official MCP Registry as global.mycelium/emissions-database, so clients that browse the registry can find us there too.
02 | Get Connected
One URL, any MCP client
Everything runs through a single streamable HTTP endpoint. There is no authentication step: paste the URL into any MCP-compatible client and you're connected.
https://mycelium.global/mcp/server
Claude
Remote server · No auth · Read-only
- Open Settings → Connectors
- Choose Add custom connector
- Name it Mycelium and paste the server URL
- Leave authentication blank and save
Custom connectors are available on paid Claude plans. Team and Enterprise admins can enable it for the whole workspace.
Claude Code
Remote server · No auth · Read-only
claude mcp add --transport http mycelium https://mycelium.global/mcp/serverRun it in any project, then ask about company emissions straight from your terminal or IDE.
ChatGPT
Remote server · No auth · Read-only
- Open Settings → Apps & Connectors
- Under Advanced settings, enable Developer mode
- Choose Create, paste the server URL
- Set authentication to None and save
Requires a paid ChatGPT plan. Enable the connector in a conversation to use the tools.
Cursor
Remote server · No auth · Read-only
Add to CursorOr add it by hand in .cursor/mcp.json:
{
"mcpServers": {
"mycelium": {
"url": "https://mycelium.global/mcp/server"
}
}
}
VS Code
Remote server · No auth · Read-only
Install in VS CodeOr add it by hand in mcp.json:
{
"servers": {
"mycelium": {
"type": "http",
"url": "https://mycelium.global/mcp/server"
}
}
}
Your own agents
Remote server · No auth · Read-only
Point any MCP client library at the endpoint over streamable HTTP, or pass it as a remote MCP server in the Anthropic or OpenAI APIs. No key management, no OAuth flow, nothing to provision.
03 | The Tools
Three tools, the whole database
Once connected, your assistant can call three read-only tools. It chains them itself: search for the company, then pull the profile or the full figures. Try "Compare Tesco and Sainsbury's Scope 1 emissions for the last reported year".
search-entities
Searches the database by company name, with an optional country filter. Returns matches with their Mycelium Score and the slug handle the other tools use.
get-entity-profile
The company's Mycelium Score and Transparency Score, its industry classification, and which reporting years have figures on file.
get-emission-figures
Complete greenhouse gas figures for one reporting year: Scope 1, 2 and 3 with categories, in the OpenCCF data model, each line tagged as reported or estimated.
04 | Access & Limits
Free, anonymous, read-only
- No API key, no account. The endpoint is open by design, so a single pasted URL works in any client.
- Read-only public data. The server exposes the same open-access data you see on any Mycelium company profile, nothing more.
- Rate limited, not gated. 30 tool calls per minute per IP, which is plenty for conversational use. Bulk extraction belongs on the API instead.
- Provenance built in. Every emissions line is tagged as reported by the company or estimated by Mycelium, and the server instructs assistants to preserve that distinction.
- Same licences as the platform. Use of the data is governed by our emissions data terms of use.
05 | Who It's For
From procurement chats to research agents
- Sustainability and procurement teams screening suppliers from inside their AI assistant.
- Carbon accountants drafting client reports with live company figures pulled into the conversation.
- Researchers and analysts running comparative studies without leaving their notebook environment.
- Software teams wiring Mycelium into custom AI agents and copilots without writing API client boilerplate.
06 | Talk to us
Building something with MCP?
If you're building an AI product on top of Mycelium, or the anonymous rate limit doesn't fit your use case, we'd love to hear from you. Higher limits and keyed access are available for serious integrations.
