I joined the team 4 weeks ago. First thing I did was build a real app with the API.

I joined Tabstack four weeks ago. The fastest way I know to understand a product is to build something real with it—not tutorials, not toy examples, but an actual app that uses the API under real conditions and breaks in interesting ways.

So I built Rival. Open-source competitive intelligence dashboard that tracks competitor pricing, changelogs, careers, docs, and GitHub signals, diffs what changes, and generates intelligence briefs automatically.

Every endpoint gets exercised for a different reason. extract/json handles structured pages with schemas I defined. extract/markdown handles changelogs—cheaper and faster for unstructured text. automate kicks in as a fallback when extraction hits a JS-heavy page. generate writes the diff summaries and intelligence briefs. /research powers Deep Dive mode, multi-source cited reports on demand.

Using every endpoint surfaced real things. Schema edge cases that don't show up in unit tests. Fallback behavior that needed work. Missing fields that only appeared on certain page types. In four weeks, building Rival has already driven a handful of real improvements and bug fixes to Tabstack—issues I never would have found reading docs.

The /insights page ended up being the most useful artifact—it shows extraction quality metrics per endpoint and surfaces the most common missing schema fields across all your scans. That's the feedback loop that makes extraction reliable over time.

MIT licensed, self-hosted, Postgres-backed. If you're building on Tabstack and want a real-world reference app that flexes everything, Rival is it.

Demo: https://rivaldemo.com/
Github Repo: https://github.com/tessak22/rival