Input and Output Schemas - Beam

Use inputs and outputs schemas to validate data on functions, endpoints, and task queues:

app.py

from beam import endpoint, schema


class Inputs(schema.Schema):
    text = schema.String()
    repeat = schema.Integer()


class Outputs(schema.Schema):
    result = schema.String()


@endpoint(name="structured", inputs=Inputs, outputs=Outputs)
def handler(inputs):
    return {"result": inputs.text * inputs.repeat}

beam deploy app.py:handler

Send {"text": "hello", "repeat": 2} to the endpoint. The handler receives a validated Inputs instance. You can also accept a context argument.

Field Types

Field	Accepted value
`schema.String()`	String
`schema.Integer()`	Integer
`schema.JSON()`	JSON-compatible input; use an object or array for serialized outputs
`schema.Object(NestedSchema)`	Object validated against a nested schema
`schema.File()`	File-like object, URL, or base64 data
`schema.Image()`	Image data validated using Pillow

All fields are required. Use Schema.new(data) to validate locally and .dump() to serialize. File outputs become public URLs. Install Pillow to use image fields. Custom validation methods and image-field limits are not preserved remotely; check those limits in your handler.