Ask your AI. It answers with your data.
The Dcycle MCP server connects your AI assistant to your account. Ask about your footprint, your facilities or your suppliers, and have it load data or build a life cycle assessment. Nothing to install, with your user and your permissions.
- 8
- compatible AI assistants
- 0
- installs: the server is remote
- 2,000+
- companies report with Dcycle
Summarise our 2025 carbon footprint by scope.
get_greenhouse_gas_emissionsReadget_organization_treeRead The 2025 footprint is 5,870 t CO₂e for the whole group. Scope 3 accounts for half, mainly from purchased goods and services.
Which facility used the most electricity last quarter?
list_facilitiesReadlist_invoicesRead Plant A, with 412 MWh: more than half of the quarter’s electricity consumption. It is the only one up on the previous quarter.
Create a category from this packaging Excel file and upload it.
create_datasetWriteadd_dataset_fieldsWritecreate_dataset_recordsWrite I’ve created “Packaging consumption” in My categories with 4 columns and uploaded the 96 rows from the file. Shall I add a column with the recycled weight?
Prepare the 2025 fleet report.
list_vehiclesReadget_greenhouse_gas_emissionsRead You have 42 vehicles: diesel accounts for 71% of fleet emissions. The three highest emitters are Plant A delivery vans.
Claude
ChatGPT
GitHub Copilot
Copilot Studio
Gemini CLI
Cursor
Windsurf
More than 2,000 European companies trust Dcycle
Your question goes to Dcycle and comes back with your data.
Your assistant doesn’t keep a copy of your data: each time it needs something, it asks Dcycle for exactly what is required, with your user and your permissions.
You ask“Which plant uses the most electricity?”
It picks the toollist_facilities · list_invoices
Reads with your userThe first time you sign in at app.dcycle.io; after that it uses a short-lived token.
Returns the dataOnly what it asked for, from your organizations.
Answers youIn words, with a table or a chart.
One address. The assistant you already use.
Add the server address to your assistant and sign in with your Dcycle account. The first time, your browser opens so you can sign in; after that, just ask.
https://mcp.dcycle.io/mcp - 1Add a custom connector.
- 2Name: Dcycle. URL: the server address.
- 3The first time, sign in to Dcycle from your browser.
- 1On the Plus, Pro, Business or Enterprise plans.
- 2Turn on developer mode in Settings, Apps, Advanced settings.
- 3Add a connector called Dcycle with the server address.
- 1Run this command in your terminal:
claude mcp add --transport http dcycle https://mcp.dcycle.io/mcp - 1In VS Code 1.101 or later, in agent mode.
- 2Add it to .vscode/mcp.json:
{ "servers": { "dcycle": { "type": "http", "url": "https://mcp.dcycle.io/mcp" } } } - 1Add it to ~/.gemini/settings.json and authenticate with /mcp auth dcycle:
{ "mcpServers": { "dcycle": { "url": "https://mcp.dcycle.io/mcp" } } } - 1With Node.js installed, add it to ~/.cursor/mcp.json:
{ "mcpServers": { "dcycle": { "command": "npx", "args": ["mcp-remote", "https://mcp.dcycle.io/mcp"] } } } It also works with Windsurf and Microsoft Copilot Studio, and with any MCP-compatible client.
Your AI sees what you see. Not one bit more.
The assistant signs in with your account and your permissions, and only receives what it asks for to answer you. Your password never goes through the AI and you can revoke access whenever you like.
Security in the documentation- OAuth 2.0 with PKCEYou sign in at app.dcycle.io, not in the assistant
- Short-lived tokenThe assistant never receives your password
- Only your organizationsThe same ones you already have access to
- Read and write labelledAnd deletions, as destructive
- Data stored in the EUOn AWS, in Ireland, encrypted with AES-256
- ISO 27001Certified security management
Dcycle doesn’t work alone. Your AI combines it with the rest.
If your assistant is also connected to your email, your folders or your spreadsheets, it can move data from one place to another in a single conversation.
- Finds this month’s electricity invoices in your email
- Reads consumption, period and facility
- Creates them in Dcycle with create_invoice
- Opens the action plan from your folder
- Separates targets and actions
- Loads them with create_decarb_target and create_decarb_action
- Checks the quarter’s footprint in Dcycle
- Summarises what has changed
- Posts it in the committee channel
- Reads the purchasing packaging sheet
- Suggests the columns
- Creates the category and uploads the rows in Dcycle
Each connector is added and authorised separately in your assistant.
It reads your data. And loads it too.
Your assistant uses the same tools across every area of Dcycle: it queries what you already have and, when you ask, creates records, categories or projects.
- Footprint and emissions
How much did we emit in 2025 by scope?
get_greenhouse_gas_emissionsget_total_impactslist_emission_factors - Facilities and invoices
Upload these electricity invoices to Plant A.
list_facilitieslist_invoicescreate_invoiceupdate_facility - Fleet
Which vehicles emit the most?
list_vehiclescreate_vehicle - Purchases and travel
Record this month’s trips.
list_purchaseslist_business_travelscreate_purchasecreate_business_travel - CSRD and double materiality
Summarise our material impacts.
list_double_materialityget_csrd_output_data - Life cycle assessment
Build the LCA for this product from its BOM.
get_lca_dashboardsearch_ecoinvent_activitiescreate_lcaadd_lca_material_block - My categories
Create a category from this Excel file.
query_datasetcreate_datasetcreate_dataset_records - Decarbonisation
Load this action plan.
get_decarb_planlist_decarb_actionscreate_decarb_targetcreate_decarb_action
Three ways to connect. Choose by what you need.
The MCP server is for working by conversation. For large, repeated uploads there is the command line, and for connecting systems, the API.
- MCP server Anyone on the team By talking to your AI assistant
- Instant questions and analysis
- One-off uploads from a file
- Combining Dcycle with other tools
- CLI (dcy) Technical profiles and consultancies Commands in the terminal
- Bulk batch uploads
- One file for many organizations
- Recurring tasks
- API The IT team Integration with code
- Connecting your ERP or other systems
- Automatic, periodic submissions
- Hands-off workflows
Dcycle, inside your AI.
Thirty seconds to see what it’s like to ask Dcycle from your assistant.
Frequently asked questions
What is the Dcycle MCP server?
It is the connection between your AI assistant and your Dcycle account, based on the Model Context Protocol, the open standard for connecting assistants to applications. With it, your assistant queries and loads Dcycle data when you ask.
Which assistants can I connect?
Claude, Claude Code, ChatGPT, GitHub Copilot in VS Code, Microsoft Copilot Studio, Gemini CLI, Cursor and Windsurf, and any other MCP-compatible client.
Do I have to install anything?
No. The server is remote: you add the address https://mcp.dcycle.io/mcp to your assistant and sign in with your Dcycle account. If you prefer, it can also run locally with an API key.
Can it change my data?
Yes, when you ask it to. As well as querying, it can create and update records such as invoices, purchases or vehicles, create your own categories and their rows, or build a life cycle assessment. Each tool shows whether it reads or writes, and deletions are labelled as destructive.
Is my data safe?
Access uses your Dcycle account through OAuth 2.0: you type your password at app.dcycle.io and the assistant only receives a short-lived token. It only sees the organizations you have access to.
How is it different from the API?
The API is designed for your technical team to integrate systems with code. The MCP server is designed so anyone can work with their data by talking to their AI assistant, without writing code.
See it with your own data.
Thirty minutes with our team. We show you how your data would come in, and what you would get out.
See Dcycle in action- No commitment
- With your data
- In your language
What your session looks like
- 01 You tell us what you’re being asked forCustomers, auditors or banks: what data and by when. 0–5 min
- 02 We show how your data comes inInvoices, ERP, suppliers: on an example close to yours. 5–20 min
- 03 You leave with a planWhat to automate first and how much time you save. 20–30 min
With someone from the team who then supports you