The Kavel MCP server (kavel-mcp) is used to bring Kavel's (kavel.ai) AI photo and video generation capabilities into any MCP-compatible client, such as Claude, Cursor, Cline, etc. Its role is not to directly generate images or videos, but to help users choose tools, write prompts tuned for the models, and then jump to Kavel's free generator to complete the actual generation.
Purpose When a user in an MCP client requests a certain image or video effect, the server browses the available AI photo and video generators, provides prompts adapted to the models, and opens the corresponding tool link so the user can generate on www.kavel.ai.
Main capabilities
- Browse Kavel's AI photo and video generators.
- Get model-tuned prompts.
- Open the correct generation tool.
- No account or API key is required to use the server; it is only responsible for selecting tools and writing prompts, after which it is handled by Kavel's free-tier generator.
Featured generators
- AI Hairstyle Changer
- AI Figurine Generator
- AI Pet Portrait Generator
- AI Muscle Generator
- 90s Yearbook Photos
- AI Dance Video Generator
- HD Photo Converter
How to use It can be run directly via npx. In Claude Desktop / Claude Code, point the client to the command npx -y kavel-mcp (stdio transport) and add it to the MCP configuration (claude_desktop_config.json or .mcp.json). The repository also includes the kavel-image-studio agent skill (located at skills/kavel-image-studio/), which can help select the correct Kavel generator, write model-tuned prompts, and provide generation links, and supports installation via the open agent-skills CLI, or as a Claude Code plugin marketplace installation.
Use cases Suitable for users who want to directly find Kavel image and video generation effects in an MCP client, get prompts, and jump to generation links.
Limitations The server does not directly complete image or video generation, but instead hands it over to Kavel's (www.kavel.ai) free-tier generator; using the server itself requires no account or API key. Other details about fees, pricing, and platform support are not yet provided.
