Database REST API: Turn Any SQL Query into an Endpoint

Database REST API · the hub

Turn any SQL query into a REST API endpoint

Write a query in Query Builder against any of the 59 connectors, click New endpoint, and you have a URL at api.querystreams.com with its own key. The SQL runs read-only on the Agent that already sits with your data; callers get JSON or CSV and never see the SQL, the schema or a credential. No server to host, no port to open, and every plan including Free has it. This page is the overview; each connector has its own guide below.

About 90 seconds from a saved query No server to host Data stays on your network One key per caller, revoke anytime

What you’ll end up with

A URL that returns the rows of a saved query to anyone holding a key, tested from the console in the portal before you share it. Like this:

{
  "city": "Seattle"
}

curl -X POST "https://api.querystreams.com/v1/endpoints/8f2c4d1e-6b3a-4f7e-9c21-5d0a7b3e2f19" \
  -H "X-API-Key: qsapi_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"city": "Seattle"}'

{
  "columns": [ { "name": "customer_id", "type": "integer" }, { "name": "name", "type": "string" }, { "name": "city", "type": "string" } ],
  "rows": [
    { "customer_id": 1042, "name": "Northwind Traders", "city": "Seattle" },
    { "customer_id": 1057, "name": "Alpine Outfitters", "city": "Denver" },
    { "customer_id": 1063, "name": "Harbour Foods", "city": "Halifax" }
  ],
  "meta": { "row_count": 3, "agent_duration_ms": 41, "request_id": "req_7Hq2mKd9" }
}

Most tools that put a REST API on a database, PostgREST, Hasura, DreamFactory, Directus and the frameworks you would write yourself, expose the schema: every table becomes an endpoint and the caller browses. Query Streams does the opposite. One saved query defines exactly what an endpoint returns. The caller gets those rows and nothing else about how they were produced, and you never run an API server, because the Agent you already have runs the query and api.querystreams.com does the HTTP.

What you’ll need

A Query Streams account

Every plan has the REST API. Free gives you one endpoint, two keys and 250 rows per call, which is enough to build this guide. Sign up at my.querystreams.com.

The Agent on a computer that can reach your own database, with the connection added

Often the database server itself or an always-on computer on the same network.

A saved query in Query Builder

The endpoint wraps exactly one saved query. Write the SELECT, run it once, save it. Add a Dynamic Filter (a @variable in the WHERE clause) if callers should be able to filter.

Set it up in four steps

1. Save the query

   In Query Builder pick a connection, write the SELECT and run it once. A @variable in the WHERE clause becomes an API parameter:

   Saved query · with one Dynamic Filter

   SELECT customer_id, name, city
   FROM customers
   WHERE city = @city
   ORDER BY name;

   Click Save; the name becomes the endpoint’s display name.

2. Create the endpoint

   Open API in the top bar, click New endpoint, and on Pick query click the query you saved. Click Next.

   Open API

3. Configure and lock parameters

   On Configure keep Lifespan Permanent, Response mode Static and Inner format JSON for a first endpoint; mode and format are fixed once created. On Lock params leave city as Callable, or Lock it to a fixed value the caller cannot change. Click Next.

4. Share and get your key

   Invite People by email, add a Service key for a tool, or skip for now, then click Create & get my key. Copy the 49-character key starting qsapi_; it can be revealed again later from the Sharing tab. Click Send in the console to make the first call.

Call it from anywhere

Replace the endpoint id with yours and qsapi_YOUR_KEY with your key:

curl "https://api.querystreams.com/v1/endpoints/8f2c4d1e-6b3a-4f7e-9c21-5d0a7b3e2f19?city=Seattle" \
  -H "X-API-Key: qsapi_YOUR_KEY"

{
  "columns": [ { "name": "customer_id", "type": "integer" }, { "name": "name", "type": "string" }, { "name": "city", "type": "string" } ],
  "rows": [
    { "customer_id": 1042, "name": "Northwind Traders", "city": "Seattle" },
    { "customer_id": 1057, "name": "Alpine Outfitters", "city": "Denver" },
    { "customer_id": 1063, "name": "Harbour Foods", "city": "Halifax" }
  ],
  "meta": { "row_count": 3, "agent_duration_ms": 41, "request_id": "req_7Hq2mKd9" }
}

Rows come back as objects keyed by column name, with the column list and a meta block that tells you how many rows there are, whether the result was clipped, and the request id you can quote to support. The same call as a POST puts the parameters in a JSON body, which is the right choice for long lists of values:

curl -X POST "https://api.querystreams.com/v1/endpoints/8f2c4d1e-6b3a-4f7e-9c21-5d0a7b3e2f19" \
  -H "X-API-Key: qsapi_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"params": {"city": "Seattle"}}'

?format=csv

CSV instead of JSON, header row included. Accept: text/csv does the same. Streaming endpoints return NDJSON (one JSON object per line) with a final {"_summary": true, "row_count": N} line; check that last line.

?sample=true

At most five rows, not billed. Good for wiring up a client.

?compression=lz4

An LZ4-frame body for high-volume callers with a decoder; the Code tab has the decoder snippet per language. Plain gzip happens on its own when your client sends Accept-Encoding: gzip.

Authorization: Bearer qsapi_…

Accepted as an alias for X-API-Key. Use it from browser code.

Where it plugs in

Guides by connector

Every connector Query Streams supports can be published this way. Pick yours:

Databases you run

• Microsoft SQL Server
• PostgreSQL
• MySQL
• MariaDB
• Oracle
• SQLite
• Microsoft Access
• DuckDB
• Supabase

Data warehouses

• Snowflake
• Google BigQuery

Hosted databases

• Neon
• Amazon RDS
• Azure Database for PostgreSQL
• Azure Database for MySQL
• Azure SQL
• Google Cloud SQL
• Google AlloyDB
• DigitalOcean Managed Databases
• Aiven
• Crunchy Bridge
• Heroku Postgres
• Render
• Railway
• Fly.io Postgres
• Vultr Managed Databases
• OVHcloud Managed Databases
• Scaleway Managed Databases
• OCI Database with PostgreSQL
• MySQL HeatWave
• IBM Cloud Databases for PostgreSQL
• MariaDB SkySQL
• Xata
• Timescale Cloud
• PlanetScale Postgres
• Prisma Postgres
• Koyeb
• Nile

Folders of files and logs

• File System
• CSV Folder
• Parquet Folder
• Excel Folder
• SQLite Folder
• JSONL Folder
• Multi-record CSV
• Machine Logs
• Web Server Logs
• Container Logs
• Application Logs
• System Logs
• Custom Logs

SaaS and API connectors

• Stripe
• Shopify
• HubSpot
• Google Ads
• Google Analytics 4
• Google Search Console
• ShipStation
• iTick

Consumer-side guides: Power BI and Excel via Power Query M.

What the plans allow

Every call also has a fixed 60-second query timeout and an 8 MiB response cap, and the bytes it returns count against your plan’s monthly data allowance, the same allowance the Excel and Google Sheets add-ins use. Free has one endpoint and 5 MB a month, which comfortably covers a small table called many times a day. See the pricing page for the allowances by plan.

Keeping it safe

When something doesn’t work

Every error body has the shape {"error": {"code": "…", "message": "…"}} and every response carries an X-QS-Request-Id header. Quote the id when you write to support.

Questions people ask

Related guides

• Power BI REST API: connect live SQL data without ODBC drivers
• Connect Query Streams to Claude via MCP
• Share live data, not credentials
• Database REST API feature overview