Python SDK

Installation

pip install framelane

Quick start

Upload local media with client.uploads.upload_file(...) first, then pass the returned source_url below. It picks its route by size: files up to 256 KB go inline in one request (the only route available when outbound access to storage.googleapis.com is blocked), larger ones go straight to storage in resumable chunks, each retried on its own, so a dropped connection costs one chunk rather than the file. See Uploading media.

import os
import framelane

client = framelane.Framelane(api_key=os.environ["FRAMELANE_API_KEY"])

# Submit a render
render = client.renders.create(
    width=1920,
    height=1080,
    duration=10.0,
    frame_rate=30,
    output_format="mp4",
    elements=[
        {
            "type": "video",
            "id": "clip1",
            "source_url": os.environ["SOURCE_URL"],
            "time": 0,
            "in_point": 0,
            "out_point": 10,
        }
    ],
)

# Wait for completion
completed = client.renders.wait_for_completion(render.id)
print(completed.output.url)

API coverage

Resource	Methods
`client.projects`	`create`, `get`, `list`, `delete`, `apply_ops`, `preview`, `render`, `list_renders`
`client.renders`	`create`, `get`, `list`, `cancel`, `download`, `download_url`, `poster_url`, `wait_for_completion`
`client.tasks`	`create` (+ `transcribe`), `get`, `list`, `cancel`, `download`, `wait_for_completion`
`client.previews`	`create`
`client.uploads`	`create`, `upload_file`
`client.webhooks`	`create`, `get`, `list`, `update`, `delete`, `test`, `rotate_secret`, `list_deliveries`, `retry_delivery`
`client.api_keys`	`create`, `list`, `revoke`
`client.workspace`	`get`, `get_usage`, `update`, `delete`, `list_assets`

The project loop is the recommended path for agents: create → apply_ops → preview (free, non-billable) → render (the single billable step). AsyncFramelane exposes the same resources with await. renders.create, projects.render and projects.create return a job (or project) plus an advisory ok / violations lint of the composition — RenderSubmitOut and ProjectCreated. The render is already dispatched and the project already stored when you read them, so they are a receipt, not a gate; gate on the free previews.create(render_request=..., dry_run=True) instead. Job reads (renders.get / list / cancel) return a plain JobOut with no lint. Iterate every item across all pages with framelane.auto_paginate (or auto_paginate_async):

for render in framelane.auto_paginate(client.renders.list, status="completed"):
    print(render.id)

Webhook signature verification

from framelane import verify_webhook_signature
from flask import Flask, request

app = Flask(__name__)

@app.post("/hooks/framelane")
def handle_webhook():
    is_valid = verify_webhook_signature(
        request.data,  # raw body bytes — do not re-serialize
        request.headers.get("X-Render-Signature"),
        request.headers.get("X-Render-Timestamp"),
        os.environ["FRAMELANE_WEBHOOK_SECRET"],
    )
    if not is_valid:
        return "Invalid signature", 400

    event = request.json
    # handle event...
    return "", 200

Async support

import asyncio
import framelane

async def main():
    client = framelane.AsyncFramelane(api_key=os.environ["FRAMELANE_API_KEY"])
    render = await client.renders.create(...)
    completed = await client.renders.wait_for_completion(render.id)
    print(completed.output.url)

asyncio.run(main())

Get Started

Renders

Render Examples

Recipes

Projects

Tasks

Webhooks

Guides

SDKs

Installation

Quick start

API coverage

Webhook signature verification

Async support

​Installation

​Quick start

​API coverage

​Webhook signature verification

​Async support

Installation

Quick start

API coverage

Webhook signature verification

Async support