Transform any pipeline into a high-performance API endpoint that supports low latency and high concurrency, ready for integration.

Each API comes with Swagger documentation for standardized usage and can be accessed as a shareable REST endpoint.

​
UI

​
Create an Endpoint

To publish a query as an endpoint:

  1. Click the publish button:

  1. Select “View endpoint”:

The dashboard displays key metrics including request count, data processed, latency, and error rate.

​
Use the Endpoint

Copy the API endpoint directly from the code block to use in your application:

Access additional options through the three-dot menu in the top right corner:

  • Share Swagger: Opens the Swagger documentation page for API exploration and testing
  • Add Parameter: Define custom parameters for your API calls
  • Duplicate: Create a copy of the endpoint
  • Discard changes: Revert parameter modifications
  • Delete: Remove the endpoint

​
Parameters

Endpoints support Jinja2 syntax for parameter templating.

Example of a parameterized endpoint:

​
Expressions

Use {{ param_name }} syntax for expressions.

Example: SELECT * FROM t LIMIT {{ limit }}

The internal Jinja engine handles parameter substitution automatically. No need to escape parameters - WHERE name = {{ name }} works correctly for string values.

Supported data types:

  • String
  • Int
  • Float
  • Boolean
  • Date
  • Datetime

For other data types, use String and parse within your SQL. For example, create an array from a comma-separated string:

SELECT * FROM t WHERE group IN splitByChar(',', {{ groups }})

All parameters are passed as query strings in the HTTP API:

https://api.us.airfold.co/v1/pipes/endpoint1?param1=10&param2=foo&param3=c1,c2

For additional syntax options, refer to the Jinja2 documentation.

​
Statements

Use {% if condition %}...{% endif %} syntax for conditional statements.

Example:

SELECT
{% if start_time is defined %}
    toStartOfInterval({{ start_time }}:: DateTime, INTERVAL {{ interval }} SECONDS)
{% else %}
    toStartOfInterval(now() - INTERVAL 1 DAY, INTERVAL {{ interval }} SECONDS)
{% endif %}

For additional statement options, refer to the Jinja2 documentation.

​
CLI

​
Edit YAML file

Simply add a name to the publish parameter:

insights.yaml
nodes:
  - node1:
      sql: |
        WITH
          arrayJoin(features_mentioned) AS feature
        SELECT
          count() AS num_successful_calls,
          topK(2)(feature) AS top_features
        FROM
          analysis
        WHERE
          successful = true
publish: insights

​
Push

Push this update by running:

af push insights.yaml

​
Params

In YAML params are placed in the params property:

params:
  - name: start_time
    type: datetime
  - name: end_time
    type: datetime
  - name: groups
    type: string
  - name: models
    type: string
  - name: interval
    type: int
    default: 300

Params can be used in endpoints and draft pipes.

Params cannot be used in materialized pipes.