Kie Mcp
Generate images, video, and audio from your coding agent via kie.ai without wiring dozens of model APIs by hand.
Overview
kie-mcp is an MCP server for the Build phase that routes your agent to kie.ai’s 45+ image, 70+ video, and 20+ audio models with built-in model intelligence.
What is this MCP server?
- Access 45+ image, 70+ video, and 20+ audio generation models through one MCP server
- Model intelligence layer helps pick suitable models for each generation task
- stdio npm package kie-mcp v4.0.1 with KIE_API_KEY from kie.ai
- Optional KIE_PROJECT_ROOT saves outputs under kie/assets/raw/
- Single API surface instead of maintaining separate provider SDKs
- 45+ image models
- 70+ video models
- 20+ audio models
Community signal: 1 GitHub stars.
What problem does it solve?
Hooking each image, video, and audio provider separately burns time and makes your agent prompts inconsistent across model APIs.
Who is it for?
Indie builders who want agent-driven image, video, and audio generation inside Claude Code or Cursor during product build.
Skip if: Teams that need on-prem only generation with no third-party API or those who only need a single fixed model via a minimal custom script.
What do I get? / Deliverables
After you register kie-mcp and set KIE_API_KEY, your agent can generate and save multimodal assets from one integration.
- Generated image, video, or audio assets via agent tool calls
- Files saved under kie/assets/raw/ when KIE_PROJECT_ROOT is set
Recommended MCP Servers
Journey fit
How it compares
kie.ai MCP integration for multimodal generation, not a local Stable Diffusion skill or a design-system skill.
Common Questions / FAQ
Who is kie-mcp for?
Solo and indie builders using MCP agents who want kie.ai’s large model catalog for images, video, and audio without custom API glue.
When should I use kie-mcp?
Use it during Build when you need marketing assets, demos, or in-app media and want the agent to call kie.ai models directly.
How do I add kie-mcp to my agent?
Install the npm package kie-mcp, add it as an MCP stdio server in your agent config, and set the required KIE_API_KEY from https://kie.ai/api-key.