LeapView

Define metrics in code. Explore them everywhere.

LeapView is open-source business intelligence for teams who want their analytics in Git and their data on their own infrastructure. Define a metric once, then use it across interactive dashboards, APIs, and optional AI tools with governed access.

Get started · Documentation · Current alpha · Contributing

Executive Sales, the repository's Olist-backed showcase, in a development build. It is separate from the synthetic Sales overview starter created by the authoring CLI.

Explore the full dashboard: category breakdown and orders table

Watch a category filter update the dashboard

Recorded in the development showcase. Enter health_beauty in Category to narrow all views to health and beauty products; clear it to return to the overview.

Why LeapView?

• Review analytics like code. Version connections, transformations, semantic models, refresh pipelines, and dashboards in Git. Validate the resource graph before publishing a candidate to your instance.
• Give every view the same metrics. Reuse dimensions, relationships, and business calculations across dashboards, headless queries, and agent tools.
• Explore without rebuilding. Combine charts, maps, KPI cards, filters, and analytical tables, including matrices and pivots. The browser builder offers governed drafts, previews, and YAML export.
• Keep access consistent. Apply roles, resource grants, row filters, and column masks through the governed query layer, with OIDC sign-in and SCIM provisioning for identity management.
• Run on your infrastructure. Go serves the application, DuckDB executes queries, DuckLake manages analytical snapshots, and PostgreSQL stores control state and the DuckLake catalog. Connect supported databases, object stores, and files.

Semantic access rules can be versioned with models. Identities, attribute and role assignments, resource-grant assignments, credentials, and publication state belong to the target instance. Browser YAML export does not commit to Git. See authorization for the boundaries.

Built-in AI is optional and requires a configured provider and credentials. External clients use the deployment's OAuth-protected MCP endpoint for governed catalog, query, documentation, and dashboard-authoring tools.

Get started

Choose your starting point and what you want to have running:

• Author analytics. Get a local Docker runtime and the synthetic Sales overview starter. Install the alpha CLI, then follow the analytics development guide.
• Operate an instance. Run LeapView on your infrastructure. Follow Installation for the version-matched Compose package and external PostgreSQL/storage prerequisites.
• Contribute to LeapView. Build the application from source with the Executive Sales showcase. Follow the contributor setup, including authenticated sample-data setup.

CLI archives are published for Linux and macOS on AMD64 and ARM64. Check the installation guide for host, Docker Compose, and native credential-store requirements; package availability does not establish that every host/provider workflow is qualified.

The current server release, v0.4.0-alpha.1, is for controlled evaluation. Interfaces may change and workflows may be incomplete. The default branch and its documentation can contain changes beyond that release. Use the release notes and exact artifacts when evaluating a packaged version.

One revenue definition, several views

Start with the deterministic Sales overview project created by the CLI. Its semantic model defines revenue under spec.datasets[name=sales].metrics:

- name: revenue
  type: simple
  empty: zero
  label: Revenue
  format: currency
  agg: sum
  field: amount

Its dashboard reuses that metric in a KPI under spec.visuals:

- id: revenue
  type: kpi
  query: {type: aggregate, dimensions: [], metrics: [revenue]}
  presentation: {type: kpi, note: Synthetic revenue, tone: success}

The KPI shows the total. The monthly trend groups that revenue by month; the category chart groups it by category. All three reference revenue, keeping the calculation in the semantic model. These are excerpts; the linked files contain the complete resources, including identity and layout.

flowchart LR
  accTitle: LeapView analytics resource flow
  accDescr: Connections identify inputs for Sources. Models transform Sources, Semantic Models define business meaning, and Dashboards present it. Pipelines orchestrate refresh for a Semantic Model.
  C[Connection] --> S[Source] --> M[Model] --> SM[Semantic Model] --> D[Dashboard]
  P[Pipeline] -. refresh .-> SM

Loading

Explore further

• Build: dashboard guides, visual catalog, and connector capabilities.
• Integrate: headless BI, dbt's published relations and Parquet, and semantic-model interchange.
• Operate: self-hosting, authorization, and backup and restore.
• Understand: architecture, architecture decisions, and CLI/API/resource reference.

Contribute

After setup, use task dev for the application and task ci for the local pull-request contract. task playground opens the component and chart workbench without application services or a database. Toolchain and Nix instructions live in the contributor guide.

Report bugs and feature proposals through Issues. Report vulnerabilities privately through the security policy.

License

LeapView is available under the Apache License 2.0.