ChartDB is an open-source, web-based database diagram editor that visualizes and designs a database schema from a single SQL query, and it is built for developers, DBAs, and data teams who need entity-relationship diagrams without installing a desktop client or handing database credentials to a third-party service.

What it is

ChartDB is a TypeScript application that runs in the browser and turns a database schema into a diagram. It lives in the JavaScript and React ecosystem, with React Flow handling the interactive canvas, and it is published under the GNU Affero General Public License v3.0. The project is currently in Public Beta, distributed both as a hosted cloud application and as a self-deployable package. Instead of connecting to a database directly, ChartDB asks the user to run one query against their own database and paste the resulting JSON back into the editor.

The concrete problem it solves is the friction of getting a schema into a visual form. Conventional diagramming approaches require either a desktop installation or a live connection that demands a database password and network access. ChartDB removes both: no installation is required, no database password is required, and no account is required. It also replaces the slow manual work of documenting a schema by hand, and it replays that documentation back out as a DDL script when a migration or a handoff is needed.

Key capabilities

• Smart Query schema import: run a single query, receive the schema as JSON, paste it into the editor to render the full diagram.
• AI-powered DDL export that generates a script in the target dialect, covering moves such as MySQL to PostgreSQL or SQLite to MariaDB.
• Interactive editing and annotation, so tables, relationships, and complex structures can be adjusted and labeled inside the diagram.
• Broad engine coverage: PostgreSQL, MySQL, SQL Server, MariaDB, SQLite (including Cloudflare D1), CockroachDB, and ClickHouse.
• Custom inference server support through VITE_OPENAI_API_ENDPOINT and VITE_LLM_MODEL_NAME, with a documented local vLLM example using Qwen/Qwen2.5-32B-Instruct-AWQ.
• Docker deployment via the published image ghcr.io/chartdb/chartdb:latest, exposed on port 8080.
• Privacy controls: analytics run through Fathom and can be turned off with DISABLE_ANALYTICS=true or VITE_DISABLE_ANALYTICS=true.

Who uses it and how

• Teams documenting an existing production schema, where the diagram becomes a shared reference for onboarding and design discussions.
• Engineers planning a migration between engines, using the AI export to produce a DDL script in the destination dialect.
• DBAs and backend developers who want to inspect and annotate a schema without granting an external tool live database access.
• Self-hosters who need the editor inside their own network and deploy it as a container or as a Node build.
• Shops with an existing local model deployment, who point ChartDB at a vLLM endpoint instead of a public API.

Getting started

Use the hosted app at chartdb.io, or deploy locally. Local runs use npm install followed by npm run dev, or the Docker image ghcr.io/chartdb/chartdb:latest on port 8080; AI features require either OPENAI_API_KEY or the custom OPENAI_API_ENDPOINT and LLM_MODEL_NAME pair, and the two options must not be mixed.

How it compares

No comparable tools are named in the registry entry, so ChartDB stands alone here. Its differentiators are the AGPL-3.0 licence, the option to run entirely self-hosted from a public container image, and a data path in which the schema JSON is produced by the user's own database rather than by a vendor connection.

When to use it — and when not to

Choose ChartDB when the goal is a fast, credential-free diagram of an existing schema and a willingness to operate either a Node build or a Docker container. Self-hosters must supply that runtime themselves, and AI export depends on an external LLM, whether OpenAI or a self-run inference server. It is not the right pick for teams that need the AI migration path to work with no third-party model service at all, nor for those unwilling to accept the AGPL-3.0 licence or the project's Public Beta status.

npm install
npm run dev

npm install
npm run build

npm install
VITE_OPENAI_API_KEY=<YOUR_OPEN_AI_KEY> npm run build

docker run -e OPENAI_API_KEY=<YOUR_OPEN_AI_KEY> -p 8080:80 ghcr.io/chartdb/chartdb:latest

docker build -t chartdb .
docker run -e OPENAI_API_KEY=<YOUR_OPEN_AI_KEY> -p 8080:80 chartdb

# Build
docker build \
  --build-arg VITE_OPENAI_API_ENDPOINT=<YOUR_ENDPOINT> \
  --build-arg VITE_LLM_MODEL_NAME=<YOUR_MODEL_NAME> \
  -t chartdb .

# Run
docker run \
  -e OPENAI_API_ENDPOINT=<YOUR_ENDPOINT> \
  -e LLM_MODEL_NAME=<YOUR_MODEL_NAME> \
  -p 8080:80 chartdb

VITE_OPENAI_API_ENDPOINT=http://localhost:8000/v1
VITE_LLM_MODEL_NAME=Qwen/Qwen2.5-32B-Instruct-AWQ