Rainfrog

Model Context Protocol

The AI image and video MCP server for creative agents

Rainfrog MCP lets ChatGPT, Grok Bot, Claude, Cursor, and any compatible AI agent create and direct consistent image and video campaigns in your Rainfrog studio. Add one public link as a custom MCP server, sign in with OAuth, and start creating. No API key needed.

Rainfrog MCP server URL
https://mcp.rainfrog.ai/mcp

What is the Rainfrog MCP server?

Rainfrog MCP is a remote Model Context Protocol server for visual campaign production. MCP is the open standard AI assistants use to discover and call tools from other products. Once connected, your agent can work with Rainfrog projects, creative libraries, campaign generators, canvas workflows, and finished assets from an ordinary conversation.

One endpoint

The same public Streamable HTTP URL works in every supported client.

Your account

OAuth limits every request to the Rainfrog account you authorize.

Real projects

Results remain editable and organized in your Rainfrog studio.

How do I connect an AI agent to Rainfrog MCP?

Add https://mcp.rainfrog.ai/mcp as a remote Streamable HTTP MCP server, then approve the Rainfrog OAuth sign-in. The link is public and is not a secret.

  1. 01

    Copy the endpoint

    https://mcp.rainfrog.ai/mcp

  2. 02

    Add a remote server

    Name it Rainfrog and choose Streamable HTTP when asked.

  3. 03

    Sign in with OAuth

    Approve access to the Rainfrog account you want the agent to use.

  4. 04

    Confirm the tools

    Ask the agent to list your Rainfrog projects or account details.

ChatGPT

Add Rainfrog from Plugins as a custom MCP server, then sign in with your Rainfrog account.

ChatGPT setup
  1. In ChatGPT on the web, open Plugins.
  2. Select the plus button, then Add custom MCP server.
  3. Name it Rainfrog. Under Connection, paste https://mcp.rainfrog.ai/mcp.
  4. Choose OAuth, acknowledge the risk warning, and select Create as a plugin. That saves your custom connection. It does not add a Rainfrog directory listing.
  5. Sign in to your Rainfrog account when ChatGPT asks you to approve access.
  6. Start a new chat, type @, and select Rainfrog.
ChatGPT setup
Name: Rainfrog
URL: https://mcp.rainfrog.ai/mcp
Authentication: OAuth

Rainfrog OAuth sign-in; no API key.

Grok Bot

Grok Bot has no settings form for a custom server. Paste the Rainfrog URL in a Bot chat and approve the OAuth sign-in.

Grok Bot setup
  1. Open any Bot chat in Grok Bot. This is the always-on agent app, not the Grok chatbot.
  2. Send the setup message below.
  3. Check that the confirmation card names Rainfrog and shows the Rainfrog URL, then add it.
  4. Choose Authorize and sign in to Rainfrog. Do not paste a password or key into the chat.
  5. Ask the Bot to list your Rainfrog projects. The server is then available to the Bots on your account.
Grok Bot setup
Add a custom MCP server named Rainfrog at https://mcp.rainfrog.ai/mcp using Streamable HTTP. Do not paste any key, token, or password. When the server is added, start the Rainfrog OAuth sign-in.

Rainfrog OAuth sign-in; no API key.

Claude

Add Rainfrog under Settings → Connectors. The same custom URL works in Claude on the web, desktop, and mobile.

Claude setup
  1. Open Settings → Connectors in Claude.
  2. Choose Add custom connector and name it Rainfrog.
  3. Paste https://mcp.rainfrog.ai/mcp as the connector URL.
  4. Sign in to Rainfrog and approve access when Claude opens the authorization page.
  5. Ask Claude to list your Rainfrog projects.
Claude setup
https://mcp.rainfrog.ai/mcp

Rainfrog OAuth sign-in; no API key.

Cursor

Use the one-click install link or add Rainfrog to your personal or project MCP config.

Cursor setup
  1. Select Add to Cursor and approve the server, or paste the JSON into your MCP config.
  2. Complete the Rainfrog OAuth sign-in when Cursor asks.
  3. Open Agent and ask it to list your Rainfrog projects.
Cursor setup
{
  "mcpServers": {
    "rainfrog": {
      "url": "https://mcp.rainfrog.ai/mcp"
    }
  }
}

Rainfrog OAuth sign-in; no API key.

Claude Code

Register Rainfrog as a remote HTTP server with one terminal command, then complete OAuth in the browser.

  1. Run the command below in your terminal.
  2. Complete the Rainfrog sign-in in the browser tab that opens.
  3. Ask Claude Code to list your Rainfrog projects.
Claude Code setup
claude mcp add --transport http rainfrog https://mcp.rainfrog.ai/mcp

Rainfrog OAuth sign-in; no API key.

Any MCP client

Any client that supports remote Streamable HTTP MCP servers and OAuth can connect to Rainfrog.

  1. Create a remote MCP server named Rainfrog.
  2. Set its Streamable HTTP URL to https://mcp.rainfrog.ai/mcp.
  3. Start the connection and complete the Rainfrog OAuth sign-in.
  4. Ask the client to list Rainfrog projects to verify access.
Any MCP client setup
https://mcp.rainfrog.ai/mcp

OAuth 2.1 with dynamic client registration; no API key.

What can an AI agent do through Rainfrog MCP?

An agent can take a campaign from references to finished, downloadable assets while keeping the project editable on the Rainfrog canvas.

CapabilityWhat to askMCP tools
Create a campaignChoose a campaign preset and generate a consistent multi-shot set with the same character, product, style, and direction.“Create a 10-shot fashion editorial with this model and this product.”list_presets, start_campaign, get_run
Generate and revise imagesGenerate a single campaign image or art-direct one frame from a completed campaign without starting over.“Make shot 7 nighttime, with harder flash and a wider crop.”generate_image, regenerate_shot
Manage creative librariesList, create, inspect, update, and organize reusable characters, products, styles, environments, and brands.“Save these photos as a character named Luna in my Personal workspace.”list_library, create_character, create_product, list_assets, create_asset
Build and edit workflowsCreate projects, place templates, add and connect nodes, change node settings, and run image or video workflows.“Create a project, place the Fashion Shooting workflow, and connect this product.”create_project, instantiate_workflow, add_node, connect_nodes, run_workflow
Direct from an interactive studioOpen the Rainfrog canvas inside a compatible MCP app host, inspect the graph, focus nodes, and continue editing visually.“Open this project in the Rainfrog studio and focus the generator branch.”open_studio, get_project, inspect_nodes, focus_nodes
Create video branchesTurn a selected image output into an image-to-video branch, select a model, and run the resulting workflow.“Animate this selected frame as a five-second campaign video.”create_video_branch, set_video_model, run_workflow
Export finished assetsRetrieve short-lived download links for generated images and videos while keeping source media private.“Export the final four images and give me download links.”get_execution, export_assets

What can I ask a Rainfrog agent?

Describe the creative outcome in plain English. The agent discovers and combines the Rainfrog tools needed to complete it.

“Create a fashion editorial from these photos.”
“Make a studio still life of this product.”
“Turn this look into a short campaign video.”
“List my Rainfrog projects and open the latest one.”
“Create a character from these two references, then run a 6-shot lookbook.”
“Regenerate shot 2 with warmer light and share the download link.”

Is Rainfrog MCP secure?

Yes. Rainfrog MCP uses OAuth 2.1 instead of API keys. The agent can only reach the projects and libraries available to the Rainfrog account you approve.

  • No Rainfrog API key is copied into a config file or chat.
  • Revoke any client from Dashboard → Settings → Connected apps.
  • Set agent action families to Deny, Ask, or Allow in Rainfrog permissions.
  • Keep a human in the loop for generation, edits, and destructive actions.

How much does Rainfrog MCP cost?

Connecting an MCP client has no separate fee. Generation uses the same plan and credit balance as the Rainfrog web studio. The default campaign runs at Pro 1K quality for 30 credits per generated candidate; a 10-shot campaign generates 12 candidates for consistency selection and uses 360 credits.

Compare Rainfrog plans
MCP FAQ

Rainfrog MCP server FAQ

Direct answers about the endpoint, compatible clients, OAuth, security, credits, and creative tools.

What is the Rainfrog MCP server?

The Rainfrog MCP server connects AI assistants to your Rainfrog creative studio. It exposes tools for projects, reusable characters and products, campaign generation, canvas workflows, image and video generation, and asset export through the open Model Context Protocol.

What is the Rainfrog MCP server URL?

The public Streamable HTTP endpoint is https://mcp.rainfrog.ai/mcp. It is the same URL for every user and it is safe to paste into an MCP client because access is granted separately through OAuth.

Do I need a Rainfrog API key?

No. Rainfrog MCP uses OAuth 2.1. Your client opens a Rainfrog sign-in page, you approve access to your account, and the client stores its own access token. Never paste a password, token, or service key into chat.

Which AI assistants work with Rainfrog MCP?

Rainfrog works with ChatGPT, Grok Bot, Claude, Claude Code, Cursor, and any MCP client that supports remote Streamable HTTP servers with OAuth. All clients use the same Rainfrog endpoint. Add it as a custom MCP server. Rainfrog is not in a plugin directory yet.

What can an AI agent do through Rainfrog MCP?

An agent can create and organize characters and products, generate single images or multi-shot campaigns, create and edit studio projects, place workflow templates, connect canvas nodes, run image and video workflows, revise individual shots, and export finished assets.

Is Rainfrog MCP secure?

Rainfrog MCP signs you in with OAuth and scopes every request to the signed-in Rainfrog account. You can revoke connected apps in Rainfrog settings and control agent action families from the agent permissions page. Destructive and generative actions may also require confirmation in the host client.

How much does Rainfrog MCP cost?

There is no separate MCP connection fee. MCP uses the features and credits already included in your Rainfrog account. Generative actions require an active plan and consume credits at the same rates as the Rainfrog studio.

How many credits does a campaign use?

Campaign cost depends on quality and shot count. The default campaign uses Pro quality at 1K: 30 credits per generated candidate. Rainfrog generates a 20 percent overshoot for consistency selection, so a default 10-shot campaign produces 12 candidates and uses 360 credits.

How do I disconnect an AI client?

Open Rainfrog Dashboard → Settings → Connected apps and revoke the client. You can also remove Rainfrog from the client itself. Revoking the grant prevents that client from making new authenticated requests.

Custom MCP setup

Pick the client you already use

MCP registry identifier: ai.rainfrog/mcp