Create a D1 database and run SQL against it from outside Workers using the Cloudflare REST API, without deploying a Worker.

Documented steps

1. Create an API token scoped to Account > D1 > Edit (use D1 Read instead if you only need to query) per https://developers.cloudflare.com/fundamentals/api/reference/permissions/, and note your account_id from the dashboard.
2. Create the database: POST https://api.cloudflare.com/client/v4/accounts/{account_id}/d1/database with header Authorization: Bearer $TOKEN and body {"name": "my-database"} (optionally jurisdiction, primary_location_hint, or read_replication.mode). Doc: https://developers.cloudflare.com/api/resources/d1/subresources/database/methods/create/
3. Save the returned result.uuid as $DATABASE_ID from the create response.
4. Run SQL: POST /accounts/{account_id}/d1/database/{database_id}/query with body {"sql": "SELECT * FROM myTable WHERE field = ? OR field = ?;", "params": ["a","b"]}. sql supports multiple statements joined by semicolons, executed as a batch. Doc: https://developers.cloudflare.com/api/resources/d1/subresources/database/methods/query/
5. For performance-sensitive reads that don't need row objects, POST the same body shape to .../raw instead of .../query — it returns results as {columns, rows} arrays rather than row objects. Doc: https://developers.cloudflare.com/api/resources/d1/subresources/database/methods/raw/
6. To run multiple independent statements atomically in one call, POST {"batch": [{"sql":...,"params":...}, ...]} to the same /query (or /raw) endpoint.
7. Inspect each result[].meta (rows_read, rows_written, served_by_region, served_by_primary) in the response to confirm effects and track billable usage.