Skip to main content

    Draftless MCP für jede KI

    Verbinde Draftless mit deinem Workflow und erstelle Bilder, Videos, Audio, Text und 3D direkt aus deinen Prompts – in Claude, ChatGPT, Cursor, Codex, Gemini CLI und mehr.

    See the Claude, Gemini & Astra workflows

    Connect in under a minute

    Pick your client, copy the endpoint, and sign in with OAuth — no API key to manage.

    1

    Add the Draftless connector to Claude

    In Claude web or Claude Desktop: Settings → Connectors → Add custom connector. Name it “Draftless” and paste the server URL.

    Your Draftless endpoint

    2

    Connect and start creating

    Sign in with your Draftless account when the browser opens — OAuth handles everything, no API key to manage. Then ask Claude to generate an image, a video or a full faceless episode.

    Sign-in and consent happen in your browser on first connect — approved on draftless.dev/connect.

    Prefer the terminal? Switch to the for Claude.

    How does MCP work?

    “Make me a 5-second cinematic clip of ocean waves at sunset, with sound.”

    Your agent picks the right video model, quotes the credits, waits for your go-ahead, then polls the job to completion and hands back the finished MP4.

    discovergeneratejobs

    Just describe it, Draftless builds it

    One prompt in the client you already use — Draftless picks the models, quotes the cost, waits for your yes, and hands back the finished file.

    Claude · connected to Draftless

    You

    “Make me a 5-second cinematic clip of ocean waves at sunset, with sound.”

    Draftless MCP

    • recommend_pipelinepicked Veo 3.1 · 16:9 · audio on
    • execute_chainquote: 34 credits — confirm before spending?

      Approved by you — 34 credits

    • get_video_statusrendering… done

    Credit-spending tools always return a cost estimate and need an explicit confirmation before any credits are used.

    Prefer a terminal?

    The same models, credits and workflows from the command line — or bundled as a skill your coding agent can pick up on its own.

    Draftless CLI

    Log in once, then list models and prices, check your balance, run a workflow, or render one asset per row of a CSV.

    npx @draftless/cli login
    npx @draftless/cli models --type video
    npx @draftless/cli credits
    npx @draftless/cli batch product-hero --csv products.csv --out ./out

    Agent skill bundle

    Drop the Draftless skill into Claude Code or any agent that reads a skills folder. It knows the commands, and that it must quote the cost before spending.

    cp -r skills/draftless ~/.claude/skills/draftless