Using HTTP, you may check the health of your EdgeDB instance, check metrics on your instance, and make queries.

Your instance’s URL takes the form of http://<hostname>:<port>/. For queries, you will append db/<database-name>/edgeql.

Here’s how to determine your local EdgeDB instance’s HTTP server URL:

  • The hostname will be localhost

  • Find the port by running edgedb instance list. This will print a table of all EdgeDB instances on your machine, including their associated port number.

  • In most cases, database_name will be edgedb. An EdgeDB instance can contain multiple databases. On initialization, a default database called edgedb is created; all queries are executed against this database unless otherwise specified.

To determine the URL of a remote instance you have linked with the CLI, you can get both the hostname and port of the instance from the “Port” column of the edgedb instance list table (formatted as <hostname>:<port>). The same guidance on local database names applies here.

EdgeDB exposes endpoints to check for aliveness and readiness of your database instance.

Check if your instance is alive.


If your instance is alive, it will respond with a 200 status code and "OK" as the payload. Otherwise, it will respond with a 50x or a network error.

Check if your instance is ready to receive queries.


If your instance is ready, it will respond with a 200 status code and "OK" as the payload. Otherwise, it will respond with a 50x or a network error.

Retrieve instance metrics.


All EdgeDB instances expose a Prometheus-compatible endpoint available via GET request. The following metrics are made available.


Counter. Total number of compiler processes spawned.


Gauge. Current number of active compiler processes.


Counter. Total number of backend connections established.


Gauge. Current number of active backend connections.


Counter. Number of times the server could not establish a backend connection.


Histogram. Time it takes to establish a backend connection, in seconds.


Histogram. Time it takes to run a query on a backend connection, in seconds.


Counter. Total number of clients.


Gauge. Current number of active clients.


Counter. Total number of forcefully closed idle client connections.


Counter. Number of compiled/cached queries or scripts.


Histogram. Time it takes to compile an EdgeQL query or script, in seconds.


Counter. Number of unhandled errors in background server routines.

Before querying over HTTP, you must first enable the HTTP extension in your schema. Add this to your schema, outside any module:

using extension edgeql_http;

Then create a new migration and apply it using edgedb migration create and edgedb migrate, respectively.

Your instance is now able to receive EdgeQL queries over HTTP.

Enabling the HTTP extension is only required for querying over HTTP. It is not required for health checks or observability.

Make a query to your EdgeDB database using this URL:


You may make queries via either the POST or GET HTTP method. Query requests can take the following fields:

  • query - contains the EdgeQL query string

  • variables- contains a JSON object where the keys are the parameter names from the query and the values are the arguments to be used in this execution of the query.

When using the GET method, supply query and variables as query parameters. For a POST request, use the application/json content type and submit a JSON payload with query and variables as top-level keys in that payload as in this example:

Here’s an example query you might want to run to insert a new person in your database, as executed from the EdgeDB REPL:

insert Person { name := <str>$name };
Parameter <str>$name: Pat
{default::Person {id: e9009b00-8d4e-11ed-a556-c7b5bdd6cf7a}}

The query inserts a Person object. The object’s name value is parameterized in the query as $name.

This GET request would run the same query (assuming the instance is local and the database is named edgedb):

GET http://localhost:<port>/db/edgedb/edgeql?query=insert%20Person%20%7B%20name%20%3A%3D%20%3Cstr%3E$name%20%7D%3B&variables=%7B%22name%22%3A%20%22Pat%22%7D

As you can see with even this simple query, URL encoding can quickly become onerous with queries over GET.

Here’s the JSON payload of a POST request to execute the query:

  "query": "insert Person { name := <str>$name };",
  "variables": { "name": "Pat" }

The response format is the same for both methods. The body of the response is JSON of the following form:

  "data": [ ... ],
  "error": {
    "message": "Error message",
    "type": "ErrorType",
    "code": 123456

The data response field will contain the response set serialized as a JSON array.

Note that the error field will only be present if an error actually occurred. The error will further contain the message field with the error message string, the type field with the name of the type of error and the code field with an integer error code.

Caution is advised when reading decimal or bigint values using the HTTP protocol because the results are provided in JSON format. The JSON specification does not have a limit on significant digits, so a decimal or a bigint number can be losslessly represented in JSON. However, JSON decoders in many languages will read all such numbers as some kind of of 32- or 64-bit number type, which may result in errors or precision loss. If such loss is unacceptable, then consider casting the value into str and decoding it on the client side into a more appropriate type.


We use ChatGPT with additional context from our documentation to answer your questions. Not all answers will be accurate. Please join our Discord if you need more help.