I build arcodash, so read this with that in mind. It's an open-source dashboard tool (AGPL-3.0), v0.1.0 went out on Nov 8, and it's early beta.
The thing that made me start: a dashboard that's wrong because a table quietly stopped loading. Nothing errors. The chart renders, the number looks plausible, and you find out when someone asks why Tuesday looks odd.
The tool that draws the chart and the tool that watches the data are usually two different tools, set up by two different people, with two different schedules. So the watching tends to lag behind the drawing, or never gets set up.
What I did instead is put both on one scheduler. You write SQL (or ask in plain English), build a dashboard, and freshness, null rate and row count checks run next to the queries they protect. A failed check goes to Slack, and an LLM gives you a first guess at why it fired. The LLM uses your own key and can be switched off.
Why Rust: the API is Axum, and the worker is a separate binary that pulls jobs from Valkey. I wanted the scheduler and the worker to stay boring under load, and I wanted one small container image per service instead of a runtime to babysit. Postgres holds all the state. That's the whole stack, plus a React UI.
What it looks like to try:
git clone https://github.com/arcodash/arcodash.git cd arcodash cp .env.example .env docker compose up --build Then open localhost:3000. The first build compiles the Rust services, so give it a few minutes. There are 19 connectors plus a Query Results source that lets you join across your other saved queries.
It's early, so I'd like to hear what breaks on data that isn't mine.
If you don't want to run Postgres, Valkey, the worker and upgrades yourself, there's a hosted version from $59/month with 14 days free and no card: https://www.arcodash.com/?utm_source=devto&utm_medium=post&utm_campaign=growth-v2&utm_content=oss-story-devto-20261113
Repo: https://github.com/arcodash/arcodash. What would you check first on your own dashboards?
