What it does

Your agent builds the dashboard. We keep it alive.

Hosting, refresh, alerts, delivery and access control for dashboards your agent writes.

A published sales dashboard with KPI cards, a ranked bar chart and a trend line

MCP server and SDKs

Python and TypeScript. Your agent uploads data, builds the dashboard, publishes it and sets the alert, from the terminal.

A terminal installing vibedasher-mcp, then six tool calls building and publishing a dashboard

A real session, unedited.

Public dashboards

Publish to a URL. Anyone opens it with no account and no warehouse credentials.

A public dashboard of two centuries of carbon emissions

Embedded, with a signed token

Your server signs a short-lived token naming one viewer. The iframe carries it, the dashboard opens scoped to that person, and nobody signs in to us. This is the half that takes a team two quarters.

Three boxes: your app asks, your server mints a signed token for one viewer, vibedasher verifies and renders. Below, a table: who is looking, what they see, how long the token lasts, and where it runs.

Eject to code

Eject the dashboard as React source into your own repository. It keeps reading data through our API, until you decide it shouldn't.

See an ejected dashboard

A terminal running eject_viz, then the file tree it returns: App.tsx, config.ts, data.ts, a components folder, package.json and a README.

Scheduled refresh

File uploads, warehouse connections and SQL datasets, refreshed on the schedule you set.

A population dashboard refreshed on a schedule

Alerts

A threshold on any metric. Email or Slack when it is crossed, with the chart attached.

A metric crossing its threshold, delivered to email and Slack

Subscriptions

Scheduled email delivery of a dashboard, so people who never log in still get the number.

A weekly schedule and the recipients who never log in

Lineage

Datasets, dashboards and dbt models in one graph. See what breaks before you change a source.

RBAC

Users, groups, roles and folder permissions. For customers, a signed embed token carries per-viewer row filters the browser cannot widen.

A permissions matrix of groups and roles, and an embed token carrying a row filter

Row-level access

Coming soon

One dashboard, and each viewer sees only their own rows — filtered on the server, from the same signed token that lets them in.

The same groups-and-roles grid, applied per row rather than per folder.

Connectors

19 connectors, from a CSV on your laptop to Snowflake, BigQuery, Postgres and S3.

Everything else

Query lanes

wasm in the viewer's browser, or SQL on the server. Per dashboard.

dbt projects

Import models, runs and tests. They appear in lineage.

Folders

Organise dashboards and datasets, with permissions per folder.

API keys

Server-to-server access, scoped to one account.

Versioning

Every draft is a commit. Browse the history and revert a file.

Screenshots

Server-side capture of any dashboard, for a report or a channel.

Iceberg storage

Your data lands in open Parquet you can read without us.

Self-hosted

The same engine in your own network.

Six languages

The app and the docs in EN, FR, ES, DE, IT and PT-BR.

Cloud, or your own network

Vibedasher Cloud

Dashboards and data in Frankfurt. Sign-in identity is held in the United States, for every account. Pay for what you query.

Self-hosted

The same engine inside your network, so your rows never leave it. An engagement scoped with us, not a download.

How self-hosting works

Start with a dashboard you can open

Every dashboard in the showcase is live, public, and was built by an agent.

Open the showcase