Every tutorial on this blog says "connect the DODIL MCP once." This is that once — four steps, about two minutes, and you end with an agent that can drive the whole platform. Do it now and never think about it again.

1. Register (free)

Sign up at register.dodil.io — no card, no cost during alpha. You get an organization and API access on the spot.

2. Install the dodil CLI

One static binary for the whole platform — Ignite, DataK3, Models, Scriptum. On macOS or Linux:

curl -fsSL https://cli.dodil.io | bash

Then confirm it's on your PATH:

dodil version

On Windows, a different architecture, or if you'd rather grab the binary directly, the download page lists every build: dodil.io/cli.

3. Log in

dodil auth login

A short device-code flow opens your browser; approve it and the CLI is authenticated against your org. That's the whole auth story — every dodil … command now works.

dodil data health      # sanity check: {"status":"ok", ...}

4. Connect your agent (MCP)

Here's the part that makes the tutorials fun. The CLI ships an MCP server — dodil mcp — that exposes every DODIL operation (create a bucket, run SQL, embed a doc, call a model, deploy a function) as a typed tool with JSON output.

Most of the time there's nothing to do here. Once the CLI is installed and logged in, DODIL registers the MCP for you — a compatible agent discovers it out of the box. If you're on a setup that doesn't auto-discover, or you just want to add it explicitly, it's one command:

Claude Code

claude mcp add dodil -- dodil mcp

Cursor / VS Code / Codex — add the server to your MCP config:

{
  "mcpServers": {
    "dodil": { "command": "dodil", "args": ["mcp"] }
  }
}

Now your agent can see and call DODIL directly. Ask it something and watch it work:

"List my DataK3 buckets, then create one called hello-dodil."

It'll call data_bucket_list and data_bucket_create — no glue code, no copy-pasting commands.

TIP

Code mode. Add --enable-exec (claude mcp add dodil -- dodil mcp --enable-exec) to also expose a dodil_exec tool that runs a bash script with the CLI on PATH — handy when the agent wants to chain many commands in a single step.

That's it — you're building

You now have both front-ends to DODIL: the CLI for scripting, and an agent for prompting. That's why every tutorial shows each step twice — pick whichever you like, they run the same primitives.

Go build: