doru

🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧

PROJECT UNDER ACTIVE DEVELOPMENT, NOTHING WORKS, EVERYTHING CHANGES EVERY DAY UNTIL VERSION 0.1.0.

🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧🚧

A cross-platform desktop app for working with family trees: your data stays in plain folders on your disk, works fully offline, and is designed to be driven by AI agents through MCP and chat. There are no accounts, no cloud, and no telemetry — nothing leaves your machine unless you explicitly send it somewhere, and even AI features work with local models like Ollama if you prefer.

Dóru is genuinely open source: the entire app is released under the MIT license — free for any use, modification, and redistribution, with no proprietary parts and no open-core tricks.

The name comes from Proto-Indo-European *dóru — “tree”.

Highlights

• Local-first — your data stays yours. A project is a single .doru file (a SQLite database); photos and scans live in a media/ folder next to it. Nothing is hidden; everything is editable by hand or by agents. There are no accounts, no cloud, no telemetry: nothing leaves your machine unless you explicitly send it to an AI provider you configured yourself.
• Genuinely open source. The whole app is MIT-licensed: free to use, modify, and redistribute, for any purpose. What you run is what is in the repo — no source-available bait, no proprietary parts.
• AI-native. Dóru is both an MCP server — run doru --mcp-server <project> and any MCP client (Claude Desktop, Cursor) can query and edit your tree through tree_query, tree_get, tree_stats, tree_edit, sources_add, notes_write, charts_render — and an agent chat built into the UI: a step-by-step wizard sets up the provider (Ollama and LM Studio locally, custom OpenAI-compatible endpoints, or ACP agents; cloud presets return one by one), and keys are stored in the OS keychain.
• Auditable. Every mutation is recorded with its actor — user or agent (agent:{host}/{model}#{promptHash}). AI edits are always reviewable and undoable.
• Workbench-style UI. Activity bar, sidebar, tabs, editor area, panel, themes as text files, text-based settings.

Features

Genealogy

• Interactive tree canvas — the tree tab is one zoomable canvas with four charts (ancestors, descendants, hourglass, family), focus navigation with history and breadcrumbs, search-to-jump, and collapsible branches. Cards show an avatar, years, sex and line marks, and notes/sources/media badges.
• Editing on the canvas — add children, spouses and parents, rename, delete, and drag a card onto another person to reparent; the full editor (names, sex, birth/death dates with a quality marker — exact / about / before / after — plus a free-text form for dates that do not map to Gregorian) opens as a tab. Every action is one undoable step.
• Persons — the People sidebar lists everyone; click a person to focus the canvas and open their editor.
• Families — add parents, partners, and children from a person editor; chips show the members of each family, and removing someone from a family is one click.
• Sources and notes — attach a source (title and author) as a citation or reuse an existing one; keep free-form research notes on any person.
• Undo/redo — every change, including whole GEDCOM imports and edits made by agents, can be undone (Cmd+Z / Shift+Cmd+Z, Ctrl on Linux/Windows).

Import and export

• GEDCOM 5.5 / 5.5.1 import and GEDCOM 7.0 export.
• Tabs — a project opens in its own window, and entities (persons, the tree editor, settings) open as tabs inside it. Recent projects are listed on the start page, open project windows are restored on launch, and doru://open?path=… links or double-clicked *.doru files open in the running app.

AI and agents

• In-app chat — the agent reads and edits the tree with built-in tools: tree_query, tree_get, tree_stats, tree_edit, sources_add, notes_write, charts_render.
• Any provider — presets for Ollama (local), LM Studio (local), custom OpenAI-compatible endpoints, and ACP agents; a step-by-step wizard walks through the setup and fetches the model list for local providers; API keys are stored encrypted in the OS keychain.
• MCP server mode — run doru --mcp-server <project> and any MCP client (Claude Desktop, Cursor) can query and edit your tree; ACP agents are supported too.

Settings and themes

• settings.json — plain JSON with comments, edited in any editor, picked up live while the app runs.
• Themes — the interface is styled through CSS custom properties; a dark and a light theme are built in and chosen per OS appearance.
• 17 languages — the UI always follows the operating system language.

Installation

Dóru is installed through the app stores and package managers of your platform. See the installation guide for step-by-step instructions.

Documentation

See the user documentation for tree editing, AI agents, themes, and troubleshooting.

Development

Requirements: Node.js >= 22.13 (built-in node:sqlite) and npm.

npm install
npm run dev          # run the desktop app in dev mode
npm test             # unit tests (vitest)
npm run typecheck    # strict TypeScript across all workspaces
npm run build        # production build (electron-vite)
npm run e2e          # Playwright smoke test against the built app
npm run site:dev     # the VitePress website