Developers

Make videos from your code or your AI assistant.

Not live yet Make videos from your own code with the API, or let an AI assistant like Claude make them over MCP. This page shows how both will work.

Quick start

From a screen recording to a finished video in three steps.

  1. Create a key

    Make one under API keys. It's shown once, so put it straight into your server's settings.

  2. Send a recording

    Post a link to your screen recording. Handtap finds your taps and films a real hand tapping your phone.

    POST /v1/videos
  3. Get the video

    Check on it until it's done, or add a webhook and we'll tell you. You get a link to the MP4.

    GET /v1/videos/{id}

Make your first video

Start a video, then check on it until it's done. Each finished video uses 1 credit.


      

What you can send

The body of POST /videos, as JSON.

FieldWhat it does
recording_urlRequired. A link we can download your screen recording from.
device"auto" (the default) or "hand". Both film a real hand tapping your phone as it lies on a bed or desk. The recording must be upright and have taps in it; one that is landscape or has no taps fails, and failed videos are free.
hook_photo_urlOptional. A photo for the scene to copy: its bed or desk, its light, and the skin and nails of the hand.
webhook_urlOptional. Where to send video.completed or video.failed for this video.

Endpoints

All under https://api.handtap.ai/v1, with your key as a Bearer token.

EndpointWhat it does
POST /videosStarts a video. Returns {id, status: "queued", credits: 1}; the credit is only used if it finishes.
GET /videos/{id}One video: status (queued, rendering, done or failed), progress, and video_url or error.
GET /videos?limit=20Your latest videos, newest first.
GET /creditsYour plan and credits: {plan, left, used}.

Paste this into your coding agent

Gives Claude Code, Cursor or Codex what it needs to build with Handtap.