Model Context Protocol

      ClipStudios, inside your AI assistant

      Connect once and ask for the video. Your assistant writes the script, generates the footage, adds a voice-over and hands back the file — running on your ClipStudios plan, at ClipStudios prices, on EU-hosted infrastructure.

      https://mcp.clipstudios.ai/mcp

      Streamable HTTP. Clients that need SSE can use https://mcp.clipstudios.ai/sse.

      Works withClaudeClaude CodeChatGPTCursorAny MCP client

      Set up in three steps

      About a minute, and nothing to install.

      1

      Copy the connector URL

      One URL, no API key to generate and paste. It works with any client that speaks the Model Context Protocol.

      2

      Add it in your client

      In Claude: Settings → Connectors → Add custom connector. Name it ClipStudios and paste the URL. Cursor and other clients take the same URL in their MCP config file.

      3

      Sign in and create

      Your client opens a ClipStudios sign-in page. Approve it once and the tools appear — ask for a video, an image, a voice-over, a sound effect or a script in plain language.

      For clients configured from a file, the entry looks like this:

      {
        "mcpServers": {
          "clipstudios": {
            "url": "https://mcp.clipstudios.ai/mcp"
          }
        }
      }

      Eight tools, one connection

      Your assistant picks the right one from what you ask for. You never name a tool.

      ToolWhat it doesCostPlan
      generate_videoText-to-video across the model catalogue, with duration, resolution, aspect ratio and native audio.CreditsStarter
      generate_imageStills from a prompt at 1K, 2K or 4K, in the aspect ratio you ask for.CreditsStarter
      generate_voiceVoice-over from text, with voice, speed and stability controls.CreditsStarter
      generate_sound_effectSound design from a prompt, up to 22 seconds, loopable.CreditsStarter
      generate_scriptA short-form script in your format and tone, returned in the conversation.CreditsStarter
      get_jobPolls a generation until it finishes and returns the finished file's URL.FreeAny
      list_modelsThe models your plan can actually invoke, with supported durations and resolutions.FreeAny
      get_creditsYour balance, tier and reset date, so the assistant can budget before it spends.FreeAny

      The catalogue an assistant sees is filtered to your plan — Veo, Kling, Seedance, Runway and Wan among them — so it can only pick a model you are entitled to run.

      Access you can see and take back

      Giving an assistant your credits should be reversible and legible. It is.

      No keys to paste

      Connecting signs you in and provisions a key scoped to this connection. It is tagged as agent-issued, listed alongside your other keys, and revocable from the dashboard at any time.

      One credit engine

      Tools call the same public API as every other integration, so pricing, tier limits and the credit ledger are identical. Nothing about being an agent changes what a generation costs.

      A confirmation step before big spends

      Expensive requests — long durations, 1080p, native audio — come back with an estimate and wait for you to confirm instead of quietly burning credits.

      Questions

      Bring your studio into the conversation

      Plans start at €20 a month. The connector is included on every paid plan — no separate developer tier, no per-seat add-on.

      We use analytics to improve your experience. See our Privacy Policy.