Model Context Protocol

The StoryShort MCP server

Give Claude, ChatGPT and any AI agent the full StoryShort engine. Create, render and publish AI videos from a conversation, with every option of the app one tool call away.

Connect from Claude Code
claude mcp add storyshort --transport http https://api.storyshort.ai/mcp
ClaudeClaude CodeChatGPTCursorWindsurfOpenClawn8n

Full option discovery

Agents list every voice, caption theme, style, character and model, then pick instead of guessing.

Price before you spend

estimate_credits quotes the exact credit cost of a video before anything is generated.

Prompt to posted

One conversation goes from idea to a rendered MP4 published on TikTok, YouTube or Instagram.

What it looks like

Say this to Claude with the connector enabled:

“Make a 60-second faceless video about the history of coffee, neon caption style, energetic male voice. Show me the credit cost first, then post it to my TikTok.”

The agent checks the price, picks the voice from your library, creates the video, waits for the render and publishes it, asking you only when a choice is genuinely yours to make.

24 tools

Every tool ships with read-only and destructive annotations, so agent platforms know exactly what is safe to call.

Discover every option

Agents browse the full catalog before spending anything. No other AI video MCP exposes its options.

  • list_video_tools20+ video tools: faceless, UGC ads, movie maker, podcast, music video, ASMR, documentary...
  • get_video_toolThe exact option schema of one tool
  • list_voicesElevenLabs, OpenAI, Qwen voices + your clones
  • list_caption_themes33 caption styles: beast, neon, glass, comic...
  • list_image_stylesVisual styles for AI imagery
  • list_charactersPremade and custom AI characters
  • list_musicThe full background music library
  • list_brand_kitsYour brand kits: watermark, captions, outro
  • list_media_optionsMedia types and AI video tiers with per-clip credits
  • list_image_modelsStandalone image models and costs
  • list_video_modelsSora, Veo, Kling, Seedance 2.5 with capabilities
  • list_connected_accountsYour TikTok, YouTube and Instagram accounts
  • estimate_creditsExact price of a video before creating it
  • get_accountPlan and remaining credits

Create

The same pipeline as the app, driven from a conversation.

  • create_videoAny StoryShort tool with every option: media type, video tier, voice, captions theme, brand kit, references...
  • generate_scriptStoryShort's tuned script writer
  • generate_imageStandalone AI images (gpt-image, Flux, Seedream...)
  • generate_clipStandalone AI clips (Sora, Veo, Kling, Seedance 2.5)

Track

Everything is asynchronous: agents poll and keep working.

  • get_videoGeneration and render status, final URL
  • list_videosYour video library
  • get_mediaStandalone image/clip status

Publish

From prompt to posted, without leaving the chat.

  • render_videoFinal MP4 render (async)
  • publish_videoPost to TikTok, YouTube, Instagram
  • schedule_publishSchedule a post for later

Connect your agent

Claude and ChatGPT: OAuth, no key needed

Add StoryShort as a connector (URL below), log in when prompted and approve access. Tokens are scoped to your account and revocable anytime from Settings → API.

Connector URL
https://api.storyshort.ai/mcp

Cursor, Windsurf, n8n, custom clients: API key

Generate a key in Settings → API and pass it as a Bearer token:

mcpServers config
{
  "mcpServers": {
    "storyshort": {
      "type": "http",
      "url": "https://api.storyshort.ai/mcp",
      "headers": { "Authorization": "Bearer ss_your_api_key" }
    }
  }
}

Questions

Does it cost anything?

Tool calls are free. Video, image and clip generation consume the same credits as the app, and the agent can quote the exact price with estimate_credits before creating anything.

Which plans can use it?

Every plan, including Free (Free-plan limits on media types and rendering match the app). Publishing requires at least one connected social account.

Is it safe to let an agent publish?

Publishing tools are annotated as open-world actions, so well-behaved agent platforms ask for confirmation. You can also revoke a client's access at any time in Settings → API.

Where is the REST documentation?

The same capabilities are available as a REST API. See the API documentation.