Show HN: ChartDB Agent – Cursor for DB schema design

(app.chartdb.io)

124 points by guyb3 2 days ago

36 comments

View on Hacker News

Last year we launched ChartDB OSS (https://news.ycombinator.com/item?id=44972238) - an open-source tool that generates ER diagrams from your database (via query/sql/dbml) without needing direct DB access.

Now we’re launching the ChartDB Agent.

It helps you design databases from scratch or make schema changes with natural language.

You can:

- Generate schemas by simply describing them in plain English

- Brainstorm new tables, columns, and relationships with AI

- Iterate visually in a diagram (ERD)

- Deterministically export SQL script

Try it out here - https://chartdb.io/ai - no signup required.

Or sign up and use it on your own database

Would love to get your feedback :)

pella 2 days ago

"an open-source tool" --> https://github.com/chartdb/chartdb ( AGPL-3.0 license )

prompts:

https://github.com/chartdb/chartdb/blob/c3c646bf7cbb1328f4b2...

Reply View 4 replies
  • evanelias 2 days ago

    At quick glance I already see several things in that prompt are completely incorrect. For example: MariaDB has natively supported sequences for quite some time; for decades all versions of MySQL/MariaDB support "bool" or "boolean" as an alias; the timestamp default value advice is wrong as any arbitrary expression can be used; etc.

    These are all easily testable in Docker containers. There's a concerning lack of attention to detail here, or perhaps the prompt itself was also created using AI and it bakes in hallucinations from the get-go.

    As for the AGPL: in ChartDB's previous Show HN (only 6 weeks ago), I asked how they were running an enhanced paid SaaS when they had so many external contributions prior to adopting a CLA, and I did not receive a response [1].

    [1] https://news.ycombinator.com/item?id=44972986

    Reply View 3 replies
    • ehsanu1 a day ago

      I see no conflict between AGPL and SaaS: https://opensource.stackexchange.com/a/12988

      Reply View 1 reply
      • evanelias a day ago

        OP accepted a lot of third-party contributions without a CLA, and then built a paid SaaS; and as far as I can tell, they aren't offering the full source code for that paid SaaS to users. This would be a requirement to avoid infringing the copyright of the third party contributors.

        When code is contributed to a FOSS project, and there's no CLA, the contribution is licensed by the third-party contributor under the same terms as the project, and furthermore the contributor retains copyright (assuming there's no CAA, which is stronger than a CLA).

        So if you start an AGPL project and offer a paid SaaS, you need to do one of the following:

        a) don't accept outside pull requests

        b) accept outside pull requests with a CLA or CAA

        c) provide the full source code to the paid SaaS to users, as per the terms of the AGPL

        d) keep those pull requests in the AGPL repo but rewrite or remove them in the paid SaaS

        Any other path opens you up to risk of copyright infringement lawsuits if your business is successful.

        The OP eventually added a CLA, but had already accepted a lot of code contributions prior to that. So in their previous Show HN, I asked what path they took to resolve the AGPL hurdles, and they did not respond.

        Reply View 0 replies
    • politician a day ago

      > There's a concerning lack of attention to detail here.

      I choose to believe that it's a brilliant feint. The author will run all these comments back into their LLM to generate fixes for all of these issues.

      Reply View 0 replies
unboxingelf 2 hours ago

Nice thanks, I was finally able to design the schema for my dog turd tracker. The foreign key to the smells table was a great suggestion!

Reply View 0 replies
ako 2 days ago

In my experience most AI coding tools can work with databases, especially if you give it an MCP tool to connect to your (development) database. It will generate mermaid er-diagrams by reading from your database catalog, it will generate SQL to create tables, views, etc. It will generate queries, validate them and return data. It will optimize your query performance by running the query, running explain plan, looking into pg_stat_statements (or whatevery your database uses), and propose SQL optimization, indexes to be created. It will also create demo data with realistic use cases like black friday or back to school discounts for prices, etc.

A SQL MCP + mermaid is all you need.

Btw, it can also do an awesome job turning query plans into readable mermaid flow diagrams, adding details to the chart, using color to highlight bottlenecks.

Reply View 3 replies
  • all2 a day ago

    I tried my prompt with Claude and with this tool. There is no discussion with this tool. It just generates (a pretty cool) diagram.

    I'll probably stick with Claude for now.

    A killer feature would be a discussion loop that slowly refines the model in plaintext before committing to a diagram.

    Reply View 1 reply
quentindanjou 2 days ago

This is really cool! Would love to use that for quite a bit of prototyping on some personal projects!

Only thing I kinda dislike is the low readability of the connection because of that border around the table of the same color as the connection lines (I think things would look cleaner without it)

Reply View 0 replies
daxfohl 2 days ago

Wow, how can you afford offering free AI on the homepage for an OSS project? I can barely afford AI APIs even for my own personal use. Seems like it would drive costs through the roof.

Reply View 1 reply
  • nacs a day ago

    Maybe they value the training data?

    Reply View 0 replies
trjordan 2 days ago

My conversation

"Design a schema like Calendly" --> Did it

"OK let's scale this to 100m users" --> Tells me how it would. No schema change.

"Did you update the schema?" --> Updates the schema, tells me what it did.

We've been running into this EXACT failure mode with current models, and it's so irritating. Our agent plans migrations, so it's code-adjacent, but the output is a structured plan (basically: tasks, which are prompt + regex. What to do; where to do it.)

The agent really wants to talk to you about it. Claude wants to write code about it. None of the models want to communicate with the user primarily through tool use, even when (as I'm sure ChartDB is) HEAVILY prompted to do so.

I think there's still a lot of value there, but it's a bummer that we as users are going to have to remind all LLMs for a little bit to do keep using their tools beyond the 1st prompt.

Reply View 2 replies
  • skeeter2020 a day ago

    I asked it to abstract a event-specific table to a GP "events" table which it did, but kept the specific table. I asked it to delete that table and it said it did, but did not. I got stuck in a loop asking it to remove the table that the LLM insisted was not part of the schema, but was present in the diagram.

    It was easier to close the tab than fire a human, but other than that not a great experience.

    Reply View 0 replies
  • IChooseY0u 2 days ago

    isnt this what the agents are for, you assign them jobs to make changes then evaluate those changes. there is a necessary orchestration piece and maybe even a triage role to sort through things to do and errors to fix

    Reply View 0 replies
asdev 2 days ago

I can prompt an LLM to make a mermaid diagram and iterate on it. Why do I need "ChartDB"?

Reply View 2 replies
  • Aperocky 2 days ago

    I'd gonna go one step further and argue that you don't even need mermaid for creating DB schemas.. direct chat works fairly well for reference.

    Reply View 1 reply
    • ssantoshp 2 days ago

      always used direct chat with an ORM & can't complain

      Reply View 0 replies
tdhz77 2 days ago

The visual is really nice. I can appreciate this tool, but once it’s in my database it’s for the human to manage and update. Do I go back to using this tool? How can this help? I don’t want any tool to read data like mcp can. Read table names, sure.

Reply View 0 replies
daxfohl 2 days ago

It'd be nice if it also included the queries for each use case, and had a better understanding of sharding. As is, I gave it a bunch of requirements, it created some tables. Was it sufficient? No idea, it didn't say what the query was for each requirement. Would it scale? Again, no idea without the queries. (But actually, no, it wouldn't have scaled; some requirements were impossible without cross-shard scans).

Reply View 2 replies
  • encoderer 2 days ago

    99% of apps never need to shard data. Probably closer to 99.9% if they are single tenant.

    Reply View 1 reply
    • daxfohl 2 days ago

      That seems off in my experience, but maybe it's different in different industries.

      Either way, working around sharding and scale seems like a place where a good AI could help provide recommendations and tradeoffs, so it's disappointing that the model doesn't seem to deal with it well. For basic relational schema design, I don't see much benefit of having AI.

      Reply View 0 replies
stronglikedan a day ago

It seems very useful for people that haven't designed many databases (it follows good patterns), but I feel like I could build the ERD faster than I could build those detailed prompts (from the examples), and I would get the benefit of naming things my way.

Reply View 0 replies
kumarm 2 days ago

The UI presentation of Schema is really nice. Is there a javascript based UI for web that is used?

Reply View 0 replies
thomask1995 2 days ago

This is super cool!

Have you thought about making a tool to help preview/dry run migrations?

I feel that's something I would want a ton of confidence in if an LLM is making migration scripts.

Especially if it's doing scary stuff like breaking up a table.

Reply View 1 reply
  • whilenot-dev 2 days ago

    How do you expect an LLM to help here? Can't you just use OverlayFS on the database volume, or a backup thereof, and run the migration against that?

    Reply View 0 replies
EcommerceFlow 2 days ago

Super cool and useful, thank you. Was just doing some database mapping so this is perfect!

Reply View 0 replies
Macuyiko 2 days ago

On the homepage it says "Sinmple" above "Export SQL", fyi

Reply View 0 replies
uxhacker 2 days ago

And when will it do class designs? Maybe this would solve many an AI coding issue.

Reply View 0 replies
not--felix 2 days ago

The idea is great but it feels like it uses a dumb model

Reply View 0 replies
sergiotapia 2 days ago

to save tokens you should have a pre-baked replay play when you click those buttons i just want to see the end result and the process sped up since i dont know what the service does. good luck!

Reply View 0 replies
apwell23 2 days ago

what makes this an "agent"

Reply View 1 reply
  • bopbopbop7 2 days ago

    The for loop I wrote today is an Agent, at least that’s what I’m telling the investors.

    Reply View 0 replies