HStart for $8/month
Guide · MCP · Agent integrations

Connect Hatch to your AI agent over MCP

Hatch AI Studio exposes a Model Context Protocol (MCP) server, so assistants like Claude, ChatGPT and Cursor can read your studio directly. Once connected, your agent can browse your generations, open a prompt, star favourites, list canvas projects and summarise what you have spent — without you copying anything across by hand.

Server URL
https://hatchaistudio.com/mcp

The connection is secured with OAuth. Every agent signs in as you and acts strictly as your account — no other user's generations, projects or spend are reachable, and your provider API keys are never exposed over MCP.

  1. Step 1

    Copy the Hatch MCP server URL

    The Hatch AI Studio MCP server lives at https://hatchaistudio.com/mcp. You can also copy it from Settings → MCP inside the app. This one URL works for every MCP client.

  2. Step 2

    Add it as a custom connector in your agent

    In Claude, open Settings → Connectors → Add custom connector. In ChatGPT, open Settings → Connectors → Add. In Cursor or another MCP client, add a remote MCP server entry. Paste the Hatch server URL and save.

  3. Step 3

    Sign in with your Hatch account

    The connector redirects you to Hatch to sign in. Use the same email you use for the studio. If you are already signed in in that browser, this step passes straight through.

  4. Step 4

    Approve the connection

    Hatch shows a consent screen naming the client that is asking for access. Approve it and you are returned to your agent with the connection marked as connected.

  5. Step 5

    Ask your agent about your work

    Try prompts like "show my last 10 Hatch video generations", "what did I spend in Hatch this month, by model?", or "star my best image from yesterday". The agent calls the Hatch tools and answers from your real data.

What your agent can do

  • list_generations — Browse your recent image, video, audio, lip sync and LLM generations with prompts, models, output URLs and cost estimates.
  • get_generation — Open a single generation by ID to read its full prompt and output URLs.
  • star_generation — Star or unstar a generation to mark your favourites.
  • list_canvas_projects — List your canvas projects with names and last-updated times.
  • get_spend_summary — Summarise your estimated generation spend over a recent period, broken down by studio and model.

Troubleshooting

  • Sent back without connecting: you were not signed in to Hatch. Open the studio in the same browser, sign in, then add the connector again.
  • No approve screen appeared: you have already approved that client. Remove the connector in your agent and re-add it to see the consent screen again.
  • Tools return nothing: the account you signed in with has no generations yet. Run something in the studio, then retry.
  • Revoking access: remove the connector in your agent's settings. It can no longer call Hatch on your behalf.

Bring your studio into your assistant

Signed in already? The same URL and steps live in Settings → MCP inside the app.

Open Hatch AI Studio