# texting.video: for LLMs & chatbots

texting.video turns a JSON "draft" (a text-message conversation + theme +
settings) into an animated fake-texting video for TikTok, Reels, and Shorts.

You can drive it three ways:

1) Copy-paste
   Ask any chatbot to output a draft (shape below), then paste the JSON file
   into the editor: texting.video/app → import icon → choose the .json file.

2) Shareable link (no account)
   POST the draft to https://texting.video/api/drafts
     body: { "dialogue": [...], "themeId": "imessage", "settings": {...} }
     returns: { "id", "url" }  ← open url to render & export in the editor
   The video renders in the user's browser; free exports carry a small
   watermark, Pro removes it.

3) MCP (Model Context Protocol)
   Add the remote MCP server: https://texting.video/mcp  (Streamable HTTP)
   Tools: create_texting_video, list_themes, get_spec
   create_texting_video stores the draft and returns the /app?d=<id> link.

## Draft shape

{
  "themeId": "imessage",              // imessage | sms | android | teams | discord | slack | terminal | jrpg | true_crime
  "dialogue": [
    { "speaker": "Alex", "text": "ok I just sent you the address" },
    { "speaker": "Me",   "text": "wait this says it's a 3 hour drive??" },
    { "speaker": "Alex", "text": "the views are insane trust me", "reactions": ["😍"] },
    { "type": "narrator", "text": "She went anyway." }
  ],
  "settings": {
    "animationStyle": "human",        // human = typed on an on-screen keyboard | realistic | typewriter
    "showKeyboard": true,
    "contactName": "Alex"
  }
}

Notes:
- The FIRST unique speaker renders on the left (the other person). The SECOND
  unique speaker renders on the right (the protagonist). Use "Me".
- animationStyle:"human" shows messages being typed on an on-screen keyboard;
  it needs a keyboard theme (imessage, sms, or android).
- Keep messages short and punchy. 4-12 messages is ideal.

The live, machine-readable schema + a ready-to-use system prompt:
https://texting.video/api/spec