Hanzo Dashboards
A dashboard is a saved query
Views over what your own applications recorded. A widget is a query against the traces, observations and scores this service already holds — not a copy of them in a second system, and not a metrics store you have to feed on the side.
Paste and ship
Paste this into any agent. It reads the skill manifest and calls Dashboards from there.
What a dashboard is made of
Widgets, monitors, saved views and score configs — all of them queries over one store.
Widgets over your own data
A widget names what to select and how to group it. Because it queries the store that already holds your traces, a chart cannot drift from what happened.
Monitors
A saved query you want watched rather than looked at. Same definition as the widget beside it, kept in one place so the alert and the chart cannot disagree.
One store, not two
Traces, observations and scores live in the service that serves the dashboards. There is no export step, no second database to keep in sync, and nothing to reconcile after an outage.
Saved table views
The columns, filters and sort you arrived at are a preset you can name and return to. Presets belong to the project, so a colleague opens the view you meant.
Scores are yours to define
Score configs say what good means for your application — a rating, a label, a number from an evaluator. Dashboards then aggregate the thing you actually care about.
Permission per category
Access is granted by category rather than per object, and the boundary is the organization on your token. Reading somebody else's project is not a URL away.
20% to the author of the code that runs
Verify that you own a repository, and every org that deploys a project built from it pays you 20% of its metered spend, accrued each period.
Start building with Dashboards
One API key. One credit balance. Every primitive.