Integrating Dcycle: API, CLI, or MCP by data category

Most people work with Dcycle through the web app, and that is enough for the majority of cases. But Dcycle can also connect directly to your own systems so data flows in and out without anyone clicking through screens. This guide explains — in plain language — what you can and cannot connect, when it is worth it, who needs to be involved, and the risks and benefits. Everything is organised by data category (logistics, fleet, invoices, and so on).

The full technical documentation lives at code.dcycle.io. This page is the decision guide that comes first.

Do I even need to integrate?

Use the web app when your data is occasional and you enter it by hand. Consider connecting when:

• You have a lot of data to load at once (hundreds or thousands of rows).
• The same load repeats often (every day, week, or month).
• You want Dcycle to talk to another system you already use (an ERP, a fleet tool, a data warehouse).

The three ways to connect

There are three doors into Dcycle. You do not choose one forever — different teams can use different doors for different jobs.

	API	CLI	MCP
What it is	A direct connection between programs	A command tool run from a terminal	A bridge for an AI assistant
Best for	Building a custom, always-on link with another system	Loading files and automating repetitive tasks	Asking about your data — and creating records — in plain language
Who sets it up	Developers / IT	A technical person (comfortable with a terminal, but it takes more than that)	Almost anyone — you connect an AI assistant
Time to get going	Days to weeks (it is a project)	Minutes to install; hours for a script	Minutes
Ongoing maintenance	Medium — someone owns the code	Low	Very low — no code to keep
Technical support needed	High	Medium–high	Low
Main advantage	Fully tailored, real time	Fast for bulk loads and automation	No technical skills — read and create data by chatting
Main drawback	Most effort to build and maintain	Read/write, but needs a technical profile	Creates records one at a time — no bulk loading yet

These are the four questions people ask most — and which door answers each:

Can I upload a lot of data at once? Yes. Most categories take a CSV file (a simple spreadsheet export). For invoices and utility bills you can also upload the PDF and Dcycle reads it automatically. Other categories use CSV or the API instead of PDF.

Can the load happen on its own, without doing it by hand every time? Yes — that is automation. A load you run every day or week can instead happen in minutes, unattended. Your company just needs somewhere to run the automatic task (a server or scheduled job, or the system where your technical team keeps its code) plus an access key. What took a person an hour a week becomes a hands-off run.

Can I get to my data quickly without opening the web app? Yes. With the AI assistant (MCP) you just connect it — no technical skill needed. With the terminal tool (CLI) you need a technical person. Setting up takes minutes; a custom script, hours to days. If it ever breaks, your technical team (or whoever set it up) fixes it.

Can I export, filter, and keep my data tidy? Yes. You get a change history and exports, so you can track what changed — and load at the pace you need: instantly as things happen (API), in batches every day or week (CLI), or with one-off questions whenever you want (MCP).

When to use each one

The table above tells you what each door is. This tells you which to pick, what you are investing, and what you get back.

API — a permanent, automatic bridge

• Use it when you want Dcycle connected to another system you already run (an ERP, a fleet or transport tool, a data warehouse, your own app), so data flows across without anyone touching it — and when you need it in real time or at continuous volume.
• Objective: remove manual entry entirely; data appears in Dcycle the moment it is created elsewhere.
• Investment: it is a development project. It needs your IT or development team, takes days to weeks to build, and someone has to own the code afterwards. The highest effort of the three.
• Benefits: fully tailored, real time, hands-off, and it scales to any volume.
• Example: every purchase registered in your ERP is sent to Dcycle automatically, with no export step.

CLI — bulk file loads and automation, without building an app

• Use it when you load data from files (CSV or Excel) or repeat the same task by hand, and you want to save time without commissioning a custom development.
• Can it be automated? Yes — this is its strong point. You can place a CLI command inside a scheduled task (a job that runs on its own every night or week) or inside the system where your technical team already automates work.
• Cases it fits: “every Monday, take the week’s shipments CSV and upload it automatically”; “upload fleet fuel consumption at the start of each month”; “load supplier purchases nightly from an export.”
• Who is it for? A technical person. Being comfortable with the terminal is the starting point, but it takes a bit more — installing tools, handling access keys, and writing a small script. It can also be driven by an AI assistant, but the tool still has to be installed on that person’s computer and can feel more complicated than the other options.
• Objective: automate repetitive file loads without a full integration project.
• Investment: installing and signing in takes minutes; a one-off command takes minutes. An automated process (a small script plus a schedule) takes hours and is best set up by a technical profile. Ongoing maintenance is low.
• Benefits: the best effort-to-result balance for bulk and recurring loads — quick to start, and it automates without a big project.
• Example: a nightly job that uploads the day’s logistics CSV by itself.

MCP — ask (and now create) your data in plain language

• Use it when you want to work with your data by talking to an AI assistant, without opening the web app or writing code: quick questions, ad-hoc analysis, a fast report — and now also creating records (an invoice, a vehicle, a trip, a purchase) just by describing them.
• Objective: fast, conversational access to your data — reading and light data entry — with no technical friction.
• Investment: connecting the assistant takes minutes. No code and no maintenance — the cheapest option.
• Benefits: zero technical learning curve; ideal for business and sustainability profiles. It reads your data and writes individual records by chat.
• Limit: it does not yet do bulk uploads — it creates records one at a time, and some data types still cannot be created this way. For loading many rows at once, use CSV or the CLI.
• Example: ask the assistant for this quarter’s fleet emissions, or tell it “add an electricity invoice of 1,200 kWh for the Madrid office” and it creates it.

By data category

Each row shows what you can connect and how. The pros and cons of each method are in the comparison table above.

Category	How to load it	Notes
Logistics — shipments (ISO 14083 / GLEC)	CSV file, or API for a live feed	Great for automation if you ship daily
Fleet / vehicles (Scope 1)	CSV or Excel, or API	—
Facilities (offices, warehouses, plants)	CSV, API, or CLI; one by one with MCP	Few of them — any method works
Energy invoices (Scope 1 and 2)	PDF (auto-read), CSV, CLI, or API	Easiest wins — the PDF auto-read saves the most manual work
Supply chain / purchases (Scope 3.1)	CSV (web app or CLI), API; one by one with MCP	Good fit when purchases already live in an ERP
Business travel, hotels, commuting (Scope 3.6 & 3.7)	CSV (hotels & surveys), or API	—
Transport — up/downstream (Scope 3.4 & 3.9)	CSV or Excel, or API	—
Custom emission factors	CSV, via CLI	Not available via MCP
Life cycle assessment (LCA)	Excel or file, or API	Visual canvas stays in the web app

What you cannot (or should not) integrate

Risks, benefits, and who to involve

Benefits. Less manual work, fewer copy-paste errors, and fresher data because loads happen on a schedule instead of when someone remembers.

Risks to plan for.

• Access keys are like passwords. Whoever holds the key can send data — store it safely and never share it in plain text.
• Dependency on the person who built it. A custom link needs an owner. If they leave, someone has to take over.
• Changes over time. As the platform evolves, a custom integration may need small updates. The terminal tool and AI-assistant bridge absorb most of this for you.

Who to involve.

• Your IT or development team — for the API, for the terminal tool (CLI), and for automated, scheduled loads.
• Your sustainability data owner — for CSV uploads and for the AI-assistant bridge (reading, reporting, and light data entry).
• Dcycle support — to confirm what your plan allows and to point your technical team to code.dcycle.io for the setup details.