API overview

The Hyperclip API lets you programmatically run any flow you’ve built in the visual editor — scripts, image generation, video, voiceover, captions, audio mixing — without opening the Hyperclip UI. Submit a run, poll for completion, receive a finished MP4 URL. Official SDKs ship for Node and Python, or you can call the JSON-over-HTTPS endpoints directly from any language.

npm install hyperclip

pip install hyperclipco

See full SDK reference →

What you can build

Bulk video generation

Run the same flow against hundreds of inputs from a spreadsheet, CMS, or queue.

SaaS integrations

Let your users generate videos inside your own product using their Hyperclip credits.

Scheduled content

Trigger a run from a cron job, GitHub Action, or Zapier webhook.

Prompt-to-video bots

Drive a flow from Discord, Slack, or a custom chat UI.

Core concepts

One endpoint base, Bearer-token auth

All API calls go to the Hyperclip API with an Authorization: Bearer hck_live_... header. No OAuth dance. The SDKs default to the production base URL, so you only need to think about it for raw HTTP — grab the base URL from your Developers tab in the app and set it as HYPERCLIP_BASE_URL in your shell to keep curl examples on this page copy-pasteable.

Official Node + Python SDKs (or raw HTTP)

npm install hyperclip or pip install hyperclipco for typed helpers, polling utilities, and structured errors. The REST endpoints are plain JSON over HTTPS, so any HTTP client works just as well — pick whichever matches your stack. See SDKs.

Runs are asynchronous — poll for completion

POST /runs creates a run and returns immediately with an ID. Poll GET /runs/:id every few seconds until status: "completed". A run typically takes 1–5 minutes depending on how much video generation the flow contains.

Inputs override the flow's defaults per step

Flows built in the visual editor have placeholder prompts and optional media slots. The API’s inputs field lets you override any of them per run — prompts for any text_input step, media URLs for any media_input step, and per-step model options.

Same credit balance as the visual editor

Every API call deducts from the same credit balance the visual editor uses. The exact credit amount appears on the run record as credits_charged. See pricing for the full breakdown.

One final render per run

Voiceover, captions, and audio nodes don’t render inline. Instead they accumulate composition layers on each step’s artifact, and a single render at the end of the run stitches everything into one MP4 — same output as the visual editor, one render per run.

Minimal example

import { Hyperclip } from "hyperclip";

const hc = new Hyperclip();  // reads HYPERCLIP_API_KEY; base URL defaults to production

const { id } = await hc.runs.create({
  flow_id: "a3c7f1ea-...",
  inputs: { prompts: { 0: "a ferret riding a skateboard through Tokyo at night" } },
});

from hyperclipco import Hyperclip

hc = Hyperclip()  # reads HYPERCLIP_API_KEY; base URL defaults to production

ref = hc.runs.create(
    flow_id="a3c7f1ea-...",
    inputs={"prompts": {0: "a ferret riding a skateboard through Tokyo at night"}},
)

curl -X POST $HYPERCLIP_BASE_URL/runs \
  -H "Authorization: Bearer hck_live_AbCdEf123456.0123456789abcdefghijklmnopqrstuv" \
  -H "Content-Type: application/json" \
  -d '{
    "flow_id": "a3c7f1ea-...",
    "inputs": {
      "prompts": { "0": "a ferret riding a skateboard through Tokyo at night" }
    }
  }'

const res = await fetch(`${BASE}/runs`, {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.HYPERCLIP_API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    flow_id: "a3c7f1ea-...",
    inputs: { prompts: { 0: "a ferret riding a skateboard through Tokyo at night" } },
  }),
});
const { id } = await res.json();

import requests

r = requests.post(
    f"{BASE}/runs",
    headers={"Authorization": f"Bearer {HYPERCLIP_API_KEY}"},
    json={
        "flow_id": "a3c7f1ea-...",
        "inputs": {"prompts": {"0": "a ferret riding a skateboard through Tokyo at night"}},
    },
)
run_id = r.json()["id"]

The response is a run handle:

{
  "id": "6f2c1c9a-9b4e-44a8-a1e3-1d8a4f5b2c60",
  "status": "queued",
  "created_at": "2026-04-17T15:04:05Z"
}

Poll GET /runs/:id every few seconds until status is completed, then download video_url.

Where to go next

Quickstart

Step-by-step: from a fresh account to a rendered MP4 in under five minutes.

SDKs

Install, configure, and call every endpoint from Node or Python.

Authentication

How to create and revoke keys, scopes, and security best practices.

Runs endpoints

Full reference for creating, polling, listing, and cancelling runs.

Pricing

How API credits are billed, with worked examples and autorefill.

Node support

Every node type supported by the API (all 26 — full parity with the editor).

Error codes

Every error code you can see, and how to recover from each one.

Build with AI tools

Hosted MCP server for Claude, Cursor, Copilot, and other LLM-powered coding tools.

Introduction

Endpoints

Reference

Examples

AI tooling

What you can build

Bulk video generation

SaaS integrations

Scheduled content

Prompt-to-video bots

Core concepts

Minimal example

Where to go next

Quickstart

SDKs

Authentication

Runs endpoints

Pricing

Node support

Error codes

Build with AI tools

​What you can build

Bulk video generation

SaaS integrations

Scheduled content

Prompt-to-video bots

​Core concepts

​Minimal example

​Where to go next

Quickstart

SDKs

Authentication

Runs endpoints

Pricing

Node support

Error codes

Build with AI tools

What you can build

Core concepts

Minimal example

Where to go next