ERDone vs ChartDB — your schema in your repo, without running another server

ChartDB and ERDone are both browser-based ERD tools that target developers who’d rather not install a desktop application to design a database schema. They look superficially similar — drag tables on a canvas, generate SQL, share diagrams — and the buyer evaluating both should know the differences before picking one.

The shortest version: ChartDB is the more mature open-source product with AI features and live database sync. ERDone keeps your diagrams in your own GitHub repo — never on ERDone’s servers — and generates migration scripts with safety warnings before you run them. If you want AI to generate a schema from a description today and you’re willing to either pay for ChartDB’s cloud or run their server yourself, ChartDB is a strong fit. If you want your schema design checked into your repo like the rest of your code, with no infrastructure to maintain and no diagram data living on anybody’s servers, ERDone is built for that.

How they compare

The matrix is the surface. Three differences below the surface determine which tool fits which developer.

Data ownership without running a server

ChartDB’s pitch around data ownership is that it’s open-source and you can self-host. That’s true — and for some buyers, that’s exactly the right answer. For most, it’s a lot of work for a benefit ERDone delivers without any work at all.

Self-hosting ChartDB means standing up a Postgres database for ChartDB itself, running their server, configuring environment variables, exposing it on a hostname (or VPN’ing into it), keeping the host patched, handling backups, monitoring uptime, and rotating credentials. For a tool you use to design schemas. That’s a real operations cost — and it has to be paid by the same person who was hoping to spend their time designing the schema.

ERDone’s design avoids the trade. Your diagrams are stored in your own GitHub repo, committed as diagram.json and schema.sql alongside the rest of your code. ERDone’s hosted servers handle authentication, billing, and share-link snapshots (only when you generate a link). The diagram itself never lives on our servers — it lives in your repo. The free tier doesn’t even reach our servers for storage: diagrams live in your browser’s local storage. Either way, the data privacy story isn’t “trust the open-source code” — it’s “we don’t have your data to begin with.”

For the indie SaaS engineer or solo founder who values thin tools over thick infrastructure, that trade matters.

Migration scripts with safety, not just schema dumps

Both tools can generate SQL. The difference is what kind.

ChartDB is good at schema sync — point it at your database, get the current schema, edit it visually, export DDL. Useful for documentation and for greenfield design.

ERDone tracks a baseline per diagram. When you change the diagram, ERDone generates the ALTER script that takes the database from the baseline to the current state — not a full re-create, an actual migration. Before you save that migration to your repo, ERDone flags every destructive operation: dropped columns, narrowed types, NOT NULL added without a default. A summary banner tells you “3 destructive operations” and a click-through dialog makes you confirm. The buyer who’s had a DROP COLUMN sneak through a “quick schema change” once is the buyer who pays for this.

Pricing: $7 vs $25

ERDone Pro is $7/month for one user. ChartDB Pro on their cloud is $25/user/month. Both have free tiers and team plans. For a solo SaaS founder or indie dev paying out of pocket, the difference is real.

Where ChartDB wins

• Fully open-source under AGPL — you can read every line of the codebase
• AI schema generation (ER Diagram Generator + Text-to-SQL) — ERDone reviews designs with AI but doesn't generate schemas yet
• Live database sync — connect to a deployed Postgres or MySQL and reverse-engineer in seconds
• Larger community, longer in-market history, and over 22,000 GitHub stars
• Supports dialects ERDone doesn't, including BigQuery, CockroachDB, and ClickHouse

ChartDB has shipped things ERDone hasn’t. The AI features are genuinely useful for kicking off a new schema — describe what you want in prose, get a starting point. The live database sync solves a real DBA workflow that ERDone doesn’t address. And being open-source matters intrinsically to some buyers, separately from whether they self-host: they want to read the code, contribute fixes, or know the tool can outlive any company.

It’s also a more mature product with a larger community. If “actively maintained, lots of contributors, well-documented” is high on your evaluation list, ChartDB wins that.

The two tools now do different kinds of AI. ERDone ships Pair with Claude — an on-demand AI review of the schema you’ve modeled, flagging missing keys and indexes, normalization, and naming issues on Pro and Team. ChartDB’s AI runs the other direction: schema generation — describe what you want, get a starting diagram, or turn text into SQL. That’s the piece ERDone hasn’t shipped; “Describe → schema” is on the roadmap. On generation, ChartDB leads; on design review, ERDone has it today.

When to pick ChartDB over ERDone

Comparison pages that pretend the competitor is always wrong are easy to spot and easy to ignore. ChartDB is a good tool. There are buyers for whom ChartDB is the right answer, and we’d rather tell you up front than have you trial ERDone, find it doesn’t fit, and churn.

Why ERDone

ERDone is built for the developer who already has a GitHub repo, already commits SQL migrations, and wants their schema design to live in the same place as the rest of their work. If that’s you, the differences in the matrix aren’t just feature-level — they’re workflow-level.

Your diagrams live in your repo, not on our servers. There’s no infrastructure to maintain, no Postgres instance to run, no host to keep patched — just sign in with GitHub, connect a repo, and you’re working. When you change the schema, ERDone generates the migration script — not a full re-create — and flags every destructive operation before you can save it. The DROP COLUMN that would silently lose production data is the kind of mistake the tool was built to catch.

At $7/month, ERDone is priced for the indie founder and solo backend engineer paying out of pocket. The 14-day full-feature trial doesn’t ask for a card. If at the end of it ERDone isn’t right for you, your diagrams export as JSON or as SQL DDL for any of the 8 supported dialects, and your repo is still your repo. There’s nothing to disentangle.

Try ERDone

Start the 14-day trial →