cloudparts

ChartDBDormant

An open-source, browser-based ER diagram editor that builds a diagram from the JSON output of one query you run yourself, so it never needs a database password, and can export the schema as DDL for another database.

What it is
Database clients and design tools
Type
Open source
Made by
ChartDB
License
AGPL-3.0 (ChartDB Cloud is a paid hosted service)
Status
Dormant. No commits to main since April 11, 2026 (last release 1.20.1); its founders now build OneCLI, an agent harness.
Facts checked
23 October 2026
Website Source code

What it is

ChartDB draws entity-relationship diagrams in the browser. Instead of connecting to your database, it gives you a “Smart Query” for your database type; you run it in any client, and it returns the whole schema, tables, columns, indexes and foreign keys, as one JSON document that you paste into ChartDB. It supports PostgreSQL (including Supabase and Timescale), MySQL, SQL Server, MariaDB, SQLite (including Cloudflare D1), CockroachDB and ClickHouse, and can also import SQL or DBML. You then arrange tables, add annotations and areas, and edit the schema in the diagram.

The open-source app is a React single-page app built on React Flow, and it keeps diagrams in the browser through Dexie, an IndexedDB library; there is no server or account. Export writes the schema as DDL for a chosen dialect, and an AI-assisted export rewrites it for a different database, for example MySQL to PostgreSQL, using an OpenAI key or any OpenAI-compatible endpoint such as a local vLLM server. You can self-host it with one docker run. ChartDB Cloud, a hosted version launched in August 2025, adds real-time collaboration, embeddable diagrams, sync with a live database and an AI agent; it is free up to 10 tables, $25 a month for 100 tables, and $59 a month for a team of three.

Why teams pick it

  • No database credentials: you run the query, so the tool never touches the database
  • Open source, runs entirely in the browser, and self-hosts as one container
  • Export to DDL in another dialect helps plan a move between databases
  • Bring your own model, including local OpenAI-compatible servers, for the AI features

Watch out for

  • Development has slowed: no commits to main since April 2026, and its founders now work on another product
  • Still labelled public beta, and the open-source version keeps diagrams in one browser unless you export them
  • The diagram is a snapshot; keeping it in sync with the live database is a Cloud feature
  • AGPL: running a modified version as a service for others obliges you to publish your changes

Alternatives

What all database clients and design tools have in common →Compare side by side →
dbdiagram.ioCommercialA web tool that draws entity-relationship diagrams from a schema you type in DBML, an open text language, and converts between DBML and SQL for PostgreSQL, MySQL, SQL Server and Oracle.

How it fits with other technologies

Self-host ChartDB, with AI export through a local model server · bash

docker run -p 8080:80 \
  -e OPENAI_API_ENDPOINT=http://localhost:8000/v1 \
  -e LLM_MODEL_NAME=Qwen/Qwen2.5-32B-Instruct-AWQ \
  -e DISABLE_ANALYTICS=true \
  ghcr.io/chartdb/chartdb:latest

# open http://localhost:8080, pick your database, copy the Smart Query,
# run it in psql or any client, and paste the JSON it returns

See it in a real stack

No architecture uses ChartDB yet. Drop it into the playground and wire it up yourself.Open playground →
Where it came from2024–2026

Jonathan Fishner and Guy Ben-Aharon, whose careers had been in security, built ChartDB in the summer of 2024 and showed it on Hacker News on August 24, 2024 as a way to “visualize database schemas with a single query”. The single query meant users never had to hand a password to a diagram tool, and the AI-powered export, which rewrites a schema for another database, was the feature they singled out. Version 1.0.0 was released in November 2024.

After a year of open-source use they launched ChartDB Cloud in August 2025, for teams that wanted shared, embeddable diagrams kept in sync with the database, and in October 2025 the ChartDB Agent, which designs or changes schemas from plain-English requests. In 2026 they turned to a new product: while using AI agents on top of ChartDB they ran into the problem of giving agents credentials safely, and built OneCLI, which they launched as a Y Combinator Summer 2026 company in August 2026. ChartDB’s last release so far is 1.20.1, on April 7, 2026.

Why it’s built this way

Import from one query’s JSON output instead of a live connection
Asking for database credentials is a barrier, especially for production databases. A catalog query anyone can run and inspect gets the same schema without the tool ever connecting.
A browser app with local storage
No installation and no account: the diagram lives in the browser, and the same static app can be self-hosted as one container.
Use AI for cross-dialect export
Translating DDL between databases has many small differences in types and syntax; the founders used a model for it, and later added deterministic export in the Cloud version.

Timeline

  1. 2024

    Shown on Hacker News (August 24); version 1.0.0 (November 4)

  2. 2025

    ChartDB Cloud launched (August 21); ChartDB Agent (October 1)

  3. 2026

    Version 1.20.1 (April 7); the founders launch OneCLI with Y Combinator (August)

Sources5 · facts checked 23 October 2026

Improve this page on GitHub