A web-based tool for exploring JSON-stat endpoints. Fetch any JSON-stat resource — collections, bundles, or datasets — and browse its data through a rich interactive interface or via natural language using a WebMCP-enabled browser.
- Fetch any JSON-stat endpoint — enter a URL or pick from several presets (JSON-stat samples)
- Collection & bundle navigation — browse linked resources and drill into individual datasets
- Two view modes — Value List (flat) and Cross-Tabulation (pivot) with configurable dimension roles
- Dimension filtering — select/deselect individual categories from any dimension
- Column sorting — click any header to sort ascending/descending
- Full-text search — filter rows by matching text across all columns
- CSV export — download the current view (including active filters and sort order) as a CSV file
- POST requests — supply a JSON payload for endpoints that require it
- Shareable URLs — the loaded endpoint is encoded in the page URL for easy bookmarking and sharing
- Dark / light theme — toggle between themes with automatic persistence
- Responsive layout — works on desktop and mobile screens
When opened in a browser that supports WebMCP, the app registers 12 tools that let an AI model control the explorer conversationally:
| Tool | Description |
|---|---|
fetch_dataset |
Fetch a JSON-stat dataset from a URL (GET or POST) |
select_dataset |
Select a dataset from a loaded bundle or collection |
set_view_mode |
Switch between flat and pivot views |
set_dimension_role |
Assign a dimension as row or column (pivot mode) |
filter_dimension |
Filter a dimension to specific categories |
search_data |
Search across all cells in the current view |
sort_data |
Sort by any column |
export_csv |
Export current view to CSV |
get_metadata |
Retrieve dataset metadata (label, source, dimensions, …) |
get_dimensions |
List all dimensions with categories and filter state |
get_data_summary |
Get a summary of the current view (rows, mode, filters, …) |
get_preset_endpoints |
List available preset endpoints |
A screencast showing a brief demonstration of the app is available on YouTube:
The app is a static single-page application — no build step or server required. Open index.html in any modern browser, or serve it with any static HTTP server:
npx -y serve .On load, the app automatically fetches the default preset endpoint. To explore a different resource, enter a URL in the input field or select one from the Preset Endpoint dropdown.
All dependencies are loaded from CDNs at runtime:
- jsonstat-toolkit — JSON-stat processing library
- Inter — typeface (Google Fonts)
See the JSON-stat project for licensing details.