DMN docs: get started

Quickstart

Install DMN

Get the Windows installer from the download page. It installs for your user; no admin rights.
Open a project

Open a folder in the app and DMN indexes it in the background, then keeps the index current as files change. From a terminal, dmn index --watch indexes the current folder and waits until it's searchable.

Connect your agents

The app can do this for you. From a terminal:

dmn skill-init --apply   # Claude Code, Codex, OpenCode: the dmn command line
dmn mcp-init --apply     # Cursor and any other MCP client
dmn hook-init --apply    # Claude Code: brief each session with the project's notes

Leave out --apply to see what a command would change first.

Check the wiring

dmn config verify        # MCP entries, skills and hooks; read-only
dmn doctor               # the engine, GPU, models, index and MCP config

Start an agent

In the app, click + agent in the Fleet panel, pick Claude Code or Codex, and give it a task. Once it's connected, it can ask DMN instead of grepping. Or try a search yourself:
```
dmn search "where does login redirect" --curate
```

Commands

What your agents run, and what you can run yourself.

Command	What it does
`dmn search "<question>" --curate`	Ranked code sections for a question in plain words, fitted to a token budget
`dmn context <symbol>`	A symbol's definition, its callers and what a change could affect, in one call
`dmn search --mode trace <symbol>`	Who calls it
`dmn impact --symbol <symbol>`	What could break if it changes (`--path` for a file)
`dmn grep "<exact text>"`	Exact matches, shown with the code around them
`dmn repo`	Branch, status, recent commits and stashes in one call
`dmn memory list`	The project's shared notes (`dmn memory read <path>` to open one)
`dmn stats`	Tokens DMN saved your agents compared with reading whole files
`dmn bench`	A search benchmark built from your own code
`dmn projects`	The projects DMN tracks, and their index status

Agents without a shell use the same abilities through MCP tools; the engine page lists them.

In the workspace

Keys	What it does
`Ctrl`+`P`	The omnibox: commands, files, code search, panes, and handing a task to an agent
`Ctrl`+`	Flip between the fleet and the editor
`Ctrl`+`G`	The git drawer
`Ctrl`+`J`	The agent dock
`?`	Every shortcut

Click a path:line in any terminal to open that line. The spawn sheet (+ agent) starts an agent as a terminal or a chat, in the project root, a folder, or its own git worktree.

The words

Agent: A coding assistant, like Claude Code or Codex, working in a project.
Pane: A terminal or chat in the workspace that hosts one agent, or a plain shell.
Fleet: The agents working on a project right now. The Fleet panel lists them.
Workspace: A set of panes, shown as a tab. You can have several.
Project: A folder DMN indexes and serves.
Engine: DMN's background process: it indexes your projects and answers agents' questions.
Memory note: A short note an agent saved in the project's .dmn/memories, readable by every agent there.
Claim: An advisory hold an agent puts on paths it's about to edit, with an expiry. Other agents check it; it never blocks a write. Agents call it a lease.

When something's wrong

What you see	Likely cause	Fix
`dmn search` fails with no engine running	Search commands never start the engine themselves	Open the app, or run `dmn start`
An agent never asks DMN	It isn't connected, or it started before you connected it	`dmn config verify`, then `dmn skill-init --apply` or `dmn mcp-init --apply`, and restart the agent
Results miss a file you just changed	The index is still catching up	`dmn index --watch` waits until it's searchable
Anything else	—	`dmn doctor` checks the engine, GPU, models, index and MCP config, and says what to do

For AI agents

A plain-text summary of DMN and this site is at /llms.txt, for agents that read the web.

Get started