MCP Kandinsky

An MCP server for image generation through Sber's official Kandinsky API, with style choice and negative prompts

MCP serverOfficial

Low risk

We rate an entry low when it mostly gives the agent instructions and reference material.

Why this level

  • Only generates an image and saves a file into the project folder, does not change or delete existing data
  • Each call spends the API's free tier quota
All reasons and checks
Russian stack

ai-forever/mcp_kandinsky

Install

In your terminal, with SkillFoxx CLI

npx skillfoxx add mcp/mcp-kandinsky

Detects the agents on your machine, checks the risk and pins the version.

Other ways to install

Run in a terminal

claude mcp add --transport stdio --env 'KANDINSKY_API_KEY=<your KANDINSKY_API_KEY>' --env 'KANDINSKY_SECRET_KEY=<your KANDINSKY_SECRET_KEY>' mcp-kandinsky -- uvx --from mcp_kandinsky mcp-kandinsky

Or add to the file .mcp.json, in the project

{
  "mcpServers": {
    "mcp-kandinsky": {
      "command": "uvx",
      "args": [
        "--from",
        "mcp_kandinsky",
        "mcp-kandinsky"
      ],
      "env": {
        "KANDINSKY_API_KEY": "<your KANDINSKY_API_KEY>",
        "KANDINSKY_SECRET_KEY": "<your KANDINSKY_SECRET_KEY>"
      }
    }
  }
}

If the file already exists, add the server inside the mcpServers key.

Keys and settings

KANDINSKY_API_KEYsecret, required
KANDINSKY_SECRET_KEYsecret, required

Replace the values in angle brackets with your own. Keys never go into install links and are not stored by us.

Install in Cursor

The button opens the agent and offers to add the server. If nothing happens, copy the config below.

Add to the file ~/.cursor/mcp.json, for all projects

{
  "mcpServers": {
    "mcp-kandinsky": {
      "command": "uvx",
      "args": [
        "--from",
        "mcp_kandinsky",
        "mcp-kandinsky"
      ],
      "env": {
        "KANDINSKY_API_KEY": "<your KANDINSKY_API_KEY>",
        "KANDINSKY_SECRET_KEY": "<your KANDINSKY_SECRET_KEY>"
      }
    }
  }
}

If the file already exists, add the server inside the mcpServers key. For a single project, put the same block into .cursor/mcp.json.

Keys and settings

KANDINSKY_API_KEYsecret, required
KANDINSKY_SECRET_KEYsecret, required

Replace the values in angle brackets with your own. Keys never go into install links and are not stored by us.

Install in VS Code

The button opens the agent and offers to add the server. If nothing happens, copy the config below.

Run in a terminal

code --add-mcp '{"name":"mcp-kandinsky","type":"stdio","command":"uvx","args":["--from","mcp_kandinsky","mcp-kandinsky"],"env":{"KANDINSKY_API_KEY":"<your KANDINSKY_API_KEY>","KANDINSKY_SECRET_KEY":"<your KANDINSKY_SECRET_KEY>"}}'

Or add to the file .vscode/mcp.json, in the project

{
  "servers": {
    "mcp-kandinsky": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "--from",
        "mcp_kandinsky",
        "mcp-kandinsky"
      ],
      "env": {
        "KANDINSKY_API_KEY": "<your KANDINSKY_API_KEY>",
        "KANDINSKY_SECRET_KEY": "<your KANDINSKY_SECRET_KEY>"
      }
    }
  }
}

If the file already exists, add the server inside the servers key.

Keys and settings

KANDINSKY_API_KEYsecret, required
KANDINSKY_SECRET_KEYsecret, required

Replace the values in angle brackets with your own. Keys never go into install links and are not stored by us.

Run in a terminal

codex mcp add mcp-kandinsky --env 'KANDINSKY_API_KEY=<your KANDINSKY_API_KEY>' --env 'KANDINSKY_SECRET_KEY=<your KANDINSKY_SECRET_KEY>' -- uvx --from mcp_kandinsky mcp-kandinsky

Or add to the file ~/.codex/config.toml, for all projects

[mcp_servers.mcp-kandinsky]
command = "uvx"
args = ["--from", "mcp_kandinsky", "mcp-kandinsky"]
env = { KANDINSKY_API_KEY = "<your KANDINSKY_API_KEY>", KANDINSKY_SECRET_KEY = "<your KANDINSKY_SECRET_KEY>" }

If the file already exists, append the block to the end.

Keys and settings

KANDINSKY_API_KEYsecret, required
KANDINSKY_SECRET_KEYsecret, required

Replace the values in angle brackets with your own. Keys never go into install links and are not stored by us.

Add to the file ~/.gemini/settings.json, for all projects

{
  "mcpServers": {
    "mcp-kandinsky": {
      "command": "uvx",
      "args": [
        "--from",
        "mcp_kandinsky",
        "mcp-kandinsky"
      ],
      "env": {
        "KANDINSKY_API_KEY": "<your KANDINSKY_API_KEY>",
        "KANDINSKY_SECRET_KEY": "<your KANDINSKY_SECRET_KEY>"
      }
    }
  }
}

If the file already exists, add the server inside the mcpServers key.

Keys and settings

KANDINSKY_API_KEYsecret, required
KANDINSKY_SECRET_KEYsecret, required

Replace the values in angle brackets with your own. Keys never go into install links and are not stored by us.

Add to the file ~/.config/devin/mcp_config.json, for all projects

{
  "mcpServers": {
    "mcp-kandinsky": {
      "command": "uvx",
      "args": [
        "--from",
        "mcp_kandinsky",
        "mcp-kandinsky"
      ],
      "env": {
        "KANDINSKY_API_KEY": "<your KANDINSKY_API_KEY>",
        "KANDINSKY_SECRET_KEY": "<your KANDINSKY_SECRET_KEY>"
      }
    }
  }
}

If the file already exists, add the server inside the mcpServers key. Legacy Cascade keeps the MCP config in ~/.codeium/windsurf/mcp_config.json.

Keys and settings

KANDINSKY_API_KEYsecret, required
KANDINSKY_SECRET_KEYsecret, required

Replace the values in angle brackets with your own. Keys never go into install links and are not stored by us.

Formerly Windsurf.

Add to the file cline_mcp_settings.json, for all projects

{
  "mcpServers": {
    "mcp-kandinsky": {
      "command": "uvx",
      "args": [
        "--from",
        "mcp_kandinsky",
        "mcp-kandinsky"
      ],
      "env": {
        "KANDINSKY_API_KEY": "<your KANDINSKY_API_KEY>",
        "KANDINSKY_SECRET_KEY": "<your KANDINSKY_SECRET_KEY>"
      }
    }
  }
}

If the file already exists, add the server inside the mcpServers key. Open the settings file in Cline: MCP Servers tab, Configure MCP Servers.

Keys and settings

KANDINSKY_API_KEYsecret, required
KANDINSKY_SECRET_KEYsecret, required

Replace the values in angle brackets with your own. Keys never go into install links and are not stored by us.

Add to the file .roo/mcp.json, in the project

{
  "mcpServers": {
    "mcp-kandinsky": {
      "command": "uvx",
      "args": [
        "--from",
        "mcp_kandinsky",
        "mcp-kandinsky"
      ],
      "env": {
        "KANDINSKY_API_KEY": "<your KANDINSKY_API_KEY>",
        "KANDINSKY_SECRET_KEY": "<your KANDINSKY_SECRET_KEY>"
      }
    }
  }
}

If the file already exists, add the server inside the mcpServers key.

Keys and settings

KANDINSKY_API_KEYsecret, required
KANDINSKY_SECRET_KEYsecret, required

Replace the values in angle brackets with your own. Keys never go into install links and are not stored by us.

A fork of Roo Code, same .roo folders.

Add to the file opencode.json, in the project

{
  "mcp": {
    "mcp-kandinsky": {
      "type": "local",
      "command": [
        "uvx",
        "--from",
        "mcp_kandinsky",
        "mcp-kandinsky"
      ],
      "environment": {
        "KANDINSKY_API_KEY": "<your KANDINSKY_API_KEY>",
        "KANDINSKY_SECRET_KEY": "<your KANDINSKY_SECRET_KEY>"
      }
    }
  }
}

If the file already exists, add the server inside the mcp key.

Keys and settings

KANDINSKY_API_KEYsecret, required
KANDINSKY_SECRET_KEYsecret, required

Replace the values in angle brackets with your own. Keys never go into install links and are not stored by us.

Add to the file ~/.config/zed/settings.json, for all projects

{
  "context_servers": {
    "mcp-kandinsky": {
      "command": "uvx",
      "args": [
        "--from",
        "mcp_kandinsky",
        "mcp-kandinsky"
      ],
      "env": {
        "KANDINSKY_API_KEY": "<your KANDINSKY_API_KEY>",
        "KANDINSKY_SECRET_KEY": "<your KANDINSKY_SECRET_KEY>"
      }
    }
  }
}

If the file already exists, add the server inside the context_servers key.

Keys and settings

KANDINSKY_API_KEYsecret, required
KANDINSKY_SECRET_KEYsecret, required

Replace the values in angle brackets with your own. Keys never go into install links and are not stored by us.

Add to the file .codeassistant/mcp.json, in the project

{
  "mcpServers": {
    "mcp-kandinsky": {
      "command": "uvx",
      "args": [
        "--from",
        "mcp_kandinsky",
        "mcp-kandinsky"
      ],
      "env": {
        "KANDINSKY_API_KEY": "<your KANDINSKY_API_KEY>",
        "KANDINSKY_SECRET_KEY": "<your KANDINSKY_SECRET_KEY>"
      }
    }
  }
}

If the file already exists, add the server inside the mcpServers key.

Keys and settings

KANDINSKY_API_KEYsecret, required
KANDINSKY_SECRET_KEYsecret, required

Replace the values in angle brackets with your own. Keys never go into install links and are not stored by us.

You will need: uv

Checked against the repository on Sep 25, 2026, commit 6c54bb4.

Text for your agent

Add the mcp-kandinsky MCP server with uvx --from mcp_kandinsky mcp-kandinsky and the KANDINSKY_API_KEY and KANDINSKY_SECRET_KEY environment variables set to my FUSION BRAIN keys.

Other ways from the author
{
  "mcpServers": {
    "mcp-kandinsky": {
      "command": "uvx",
      "args": ["--from", "mcp_kandinsky", "mcp-kandinsky"],
      "enabled": true,
      "env": {
        "KANDINSKY_API_KEY": "ВАШ_API_KEY",
        "KANDINSKY_SECRET_KEY": "ВАШ_SECRET_KEY"
      }
    }
  }
}

Snippet from the README for .cursor/mcp.json or a similar agent config.

This is third-party code. Review the repository files before installing.

What it does

The server gives the agent one tool, kandinsky_generate_image: it takes a text prompt, an optional negative prompt, width and height, one of four styles (DEFAULT, KANDINSKY, UHD, ANIME) and saves the resulting image into a given project folder, creating a kandinsky subfolder for it. Generation runs on the official Kandinsky 3.1 API through the FUSION BRAIN site.

Who it is for. For people who need illustrations right inside their agent workflow, without switching to a separate image generation site.

Good fit when

  • You need an illustration or cover from a text prompt right inside the agent conversation
  • You need a specific image style: watercolor-like, anime or crisp UHD
  • The agent already has the project open and can save the image straight into the right folder

Not a fit when

  • You need precise repeatable compositions or editing of existing images: the tool only generates from scratch
  • The free tier's 100 requests a month is already used up and you have no paid plan

Example request

Generate a watercolor-style image of spring Moscow with lilacs and save it to the assets folder

Limitations

Requires KANDINSKY_API_KEY and KANDINSKY_SECRET_KEY from the FUSION BRAIN site. The free tier is limited to roughly 100 requests a month. The tool only generates and saves an image, there is no editing or variations. There have been no pushes for over a year.

How to disable. Remove the mcp-kandinsky block from your MCP client config.

MCP

Transport
stdio
Authentication
API key
Environment variables
Environment variables
KANDINSKY_API_KEY
required, secret
An API key issued after registering at fusionbrain.ai in the API section.
KANDINSKY_SECRET_KEY
required, secret
A secret key from the same pair.

Security check

  • Only generates an image and saves a file into the project folder, does not change or delete existing data
  • Each call spends the API's free tier quota

README in short

The Russian README describes the single generation tool, gives a parameter table, shows an example call with its response, and the JSON for connecting via uvx with two environment variables. It separately notes the free tier limit of 100 requests a month. MIT license.

FAQ

Where do I get Kandinsky keys?

After registering at fusionbrain.ai, in the API section, which also lists the free tier terms.

Where does the image get saved?

Into a kandinsky subfolder inside the given project_dir parameter, named after the filename parameter.

Editors’ pick

Skills and CLI from HeyGen: the agent writes HTML compositions and renders them into MP4 videos, captions, motion graphics and decks

SkillMedium risk53.9KRepository stars

Blender MCP

MCP for Blender

Editors’ pick

An MCP server and Blender addon: the agent creates and edits objects, materials and scenes, pulls assets and runs Python in Blender

MCP serverHigh risk29.5KRepository stars
Editors’ pick

Official Remotion skills for making videos in React: compositions, animation, captions, maps and rendering

SkillLow risk4.8KRepository stars

Agentic video studio for Claude Code, Cursor and Codex: script, voiceover, images, stock footage, music, subtitles and render

WorkflowMedium risk61.8KRepository stars
Foxx AIMCP Kandinsky

I am Foxx AI and I have already vetted this tool. Ask about install, setup or anything else, and I will keep it simple.