Installation

dtoolkit packages are published to npm under the @dtoolkit/* scope. You can install individual products based on your needs or set up the entire stack at once.

Prerequisites

Before installing any dtoolkit product, make sure your environment meets these requirements:

Requirement	Details
Node.js	v22.0.0 or later (download)
Package manager	npm (bundled with Node.js) or pnpm v9+
AI coding CLI	At least one of: Claude Code, Gemini CLI, Codex CLI, or OpenCode

Tip

You can check your Node.js version by running node --version. If you need to manage multiple versions, consider using fnm or nvm.

Install individual products

Each dtoolkit product is a standalone package. Install only what you need.

dbrain — Persistent memory server

The AI memory server with SQLite + FTS5, MCP and REST APIs, entity management, memory tiers, and federation support.

pnpm

install

pnpm add -g @dtoolkit/dbrain

npm

install

npm install -g @dtoolkit/dbrain

After installing, initialize your brain:

terminal

dbrain init

This starts an interactive wizard that creates your personal brain at ~/.dbrain/. See the dbrain product guide for full configuration options.

dcontext — AI coding CLI hooks

Injects identity, project facts, and memory into your AI coding sessions automatically. Saves transcripts before compaction.

pnpm

install

pnpm add -g @dtoolkit/dcontext

npm

install

npm install -g @dtoolkit/dcontext

After installing, initialize and install hooks for your preferred CLI:

terminal

dcontext init
dcontext install

Caution

dcontext requires a running dbrain instance. Make sure you have initialized and started dbrain before installing dcontext hooks.

dwork — AI-native project manager

Markdown-driven project management with BACKLOG.md as the source of truth, full-text search, and 21 MCP tools including code graph integration.

pnpm

install

pnpm add -g @dtoolkit/dwork

npm

install

npm install -g @dtoolkit/dwork

Initialize a project workspace:

terminal

dwork init

dops — Agent observability

Track tokens, cost, tool usage, success rates, and errors across all your AI coding sessions. Ingests transcripts from Claude Code, Codex CLI, Gemini CLI, and OpenCode.

pnpm

install

pnpm add -g @dtoolkit/dops

npm

install

npm install -g @dtoolkit/dops

Initialize and start the observability server:

terminal

dops init
dops start

dproxy — Universal CLI adapter

Invoke AI models through a unified interface across four providers. Includes a CLI for interactive use and a REST API with SSE streaming.

pnpm

install

pnpm add -g @dtoolkit/dproxy

npm

install

npm install -g @dtoolkit/dproxy

Initialize dproxy configuration:

terminal

dproxy init

Install the full stack

To install all five products at once:

pnpm

install

pnpm add -g @dtoolkit/dbrain @dtoolkit/dcontext @dtoolkit/dwork @dtoolkit/dops @dtoolkit/dproxy

npm

install

npm install -g @dtoolkit/dbrain @dtoolkit/dcontext @dtoolkit/dwork @dtoolkit/dops @dtoolkit/dproxy

Then initialize each product in order:

1. Initialize dbrain — this must come first since other products depend on it.

   terminal

   dbrain init
   dbrain start

2. Initialize dwork — set up your project workspace.

   terminal

   dwork init
   dwork start

3. Initialize dops — start observability tracking.

   terminal

   dops init
   dops start

4. Initialize dproxy — configure your model adapter.

   terminal

   dproxy init

5. Install dcontext hooks — connect everything to your AI coding CLI.

   terminal

   dcontext init
   dcontext install

Verify installation

After installing, verify each product is available and running correctly:

terminal

dbrain --version
dcontext --version
dwork --version
dops --version
dproxy --version

For the server products, check their status after starting:

terminal

dbrain status
dwork status
dops status

Product	Default ports	Status command
dbrain	REST + MCP on :7878, dashboard on :7879	dbrain status
dwork	REST + MCP on :7881, dashboard on :7882	dwork status
dops	REST + MCP on :7883, dashboard on :7884	dops status
dproxy	REST API on :7880 (when using dproxy serve)	—
dcontext	No server (hooks only)	dcontext status

Tip

Each server product exposes both MCP and REST APIs on the same port. Dashboards are served on port + 1. You can access dashboards in your browser at http://localhost:<port> (e.g., http://localhost:7879 for dbrain).

Docker

dbrain and dwork provide Docker images for containerized deployments. This is useful for running shared team brains or project management servers on remote infrastructure.

terminal

docker run -d -p 7878:7878 -p 7879:7879 -v dbrain-data:/root/.dbrain dtoolkit/dbrain
docker run -d -p 7881:7881 -p 7882:7882 -v dwork-data:/root/.dwork dtoolkit/dwork

Caution

Docker support is currently available for dbrain and dwork only. Refer to each package’s README for the latest image tags and configuration options: @dtoolkit/dbrain, @dtoolkit/dwork.

From source

If you want to contribute to dtoolkit or run the latest development version, build from source:

1. Clone the repository

   terminal

   git clone https://github.com/ivncmp/dtoolkit.git
   cd dtoolkit

2. Install dependencies

   install

   pnpm install

3. Build all packages

   Turbo handles the dependency-ordered build across the entire monorepo:

   terminal

   pnpm build

4. Run tests

   terminal

   pnpm test

5. Link packages globally (optional)

   To use the CLIs from your development checkout:

   terminal

   cd packages/dbrain && pnpm link --global
   cd packages/dcontext && pnpm link --global
   cd packages/dwork && pnpm link --global
   cd packages/dops && pnpm link --global
   cd packages/dproxy && pnpm link --global

For development workflows, each package supports a dev command that watches for changes:

terminal

pnpm --filter @dtoolkit/dbrain dev
pnpm --filter @dtoolkit/dwork dev
pnpm --filter @dtoolkit/dproxy dev

Next steps

• Getting started guide — set up your first brain and connect it to your AI coding CLI
• Product guides — deep dives into each dtoolkit product
• MCP tools reference — complete list of all available MCP tools
• CLI reference — full command documentation for every CLI