Build custom Vega/Vega-Lite dashboards in Datasette, for presentation-ready plots and visualizations.

Example Screenshot


Install this plugin in the same environment as Datasette.

$ datasette install datasette-vega-dashboards


See the example fixture in example_fixture for a minimal example. You can run the fixture via:

datasette -m metadata.json --template-dir templates/ sleep_log.db

and then open to see the dashboard.

In the minimal example, we create a canned query in metadata.json:

"read_sleep": {
  "hide_sql": true,
  "sql": "SELECT * FROM (SELECT * FROM sleep_log ORDER BY date DESC LIMIT 7) ORDER BY date",
  "write": false

The template for this canned query is extended by editing the templates/query-sleep_log-read_sleep.html file. You can use the instructions provided in the Datasette custom template documentation to determine what the name of the file should be for other resources, such as dashboards you want to display on a database or table.

Each graph should have a <script> tag wrapped in a <div> tag. The <script> tag should contain the Vega/Vega-Lite specification, and must have a type of either application/vnd.vegalite+json (for Vega-Lite) or application/vnd.vega+json (for Vega). The <div> tag will be used for rendering the actual graph, and should have whatever CSS styling you want (such as setting the width or height).

<div style="width: 1000px">
    <script type="application/vnd.vegalite+json">
    /* insert vegalite spec here */
<div style="width: 1000px">
    <script type="application/vnd.vega+json">
    /* or, insert a vega spec here */

Note that the Vega-Lite spec need not contain a data field — any data field will be replaced by the plugin based on the data returned by the canned query. For Vega specs, the data field will be prepended with a new dataset with the name data.


To set up this plugin locally, first checkout the code. Then create a new virtual environment:

cd datasette-vega-dashboards
python3 -mvenv venv
source venv/bin/activate

Now install the dependencies and test dependencies:

pip install -e '.[test]'

To run the tests:



View Github