Kie.ai MCP Server

What it does

Community MCP server and sibling CLI for Kie.ai's media API, covering image, video, music and speech models from one connector — 38 tools across image, video, audio and task utilities. Since v4 it no longer fires generations directly: the agent drafts a plan with prepare_media_generation, a human approves it, and submit_media_generation spends the credits. For filmmaker-agents that buys a spend checkpoint in front of a broad model shelf, at the cost of an extra round trip per shot.

Connect

claude mcp add kie-ai -e KIE_AI_API_KEY=YOUR_KEY -- npx -y @felores/kie-ai-mcp-server

{
  "mcpServers": {
    "kie-ai": {
      "command": "npx",
      "args": [
        "-y",
        "@felores/kie-ai-mcp-server"
      ],
      "env": {
        "KIE_AI_API_KEY": "YOUR_KEY_HERE"
      }
    }
  }
}

{
  "mcpServers": {
    "kie-ai": {
      "command": "npx",
      "args": [
        "-y",
        "@felores/kie-ai-mcp-server"
      ],
      "env": {
        "KIE_AI_API_KEY": "YOUR_KEY_HERE"
      }
    }
  }
}

Sample tools

Field notes

MIT, npm @felores/kie-ai-mcp-server still 5.0.0 (Aug 23 2026), Node >=20, 79 stars. The repo was last pushed Aug 26 2026 — Wan 3.0 support and a separate @felores/kie-ai-openai-server package (a loopback HTTP server with OpenAI-shaped image/video routes, not MCP) landed in the monorepo but not in a new MCP release, and it has been quiet for five weeks since. v4.0.0 (Aug 18) disabled direct media tool calls by default in favour of approval-bound plans: prepare_media_generation validates and creates no provider task, a host confirmation form or `kie-cli --approve <plan-id>` clears it, then submit_media_generation runs it. Set KIE_AI_ALLOW_DIRECT_GENERATION=true to restore the old one-shot behaviour, which is what unattended agents will need. v5.0.0 moved to the split SDK v2 packages and raised the floor to Node >=20 (npm engines; the release notes say the same), returns failures as isError with structuredContent, and gates official MCP Tasks behind KIE_AI_MCP_TASKS=true. Stdio by default, but v5 ships first-class HTTP: MCP_TRANSPORT=http or --http, with MCP_HTTP_HOST/MCP_HTTP_PORT and an optional KIE_MCP_HTTP_TOKEN bearer — self-hosted only via the Dockerfile and Coolify compose, there is no vendor-hosted URL. Filter the 38 tools with KIE_AI_ENABLED_TOOLS, KIE_AI_TOOL_CATEGORIES=image,video or KIE_AI_DISABLED_TOOLS (that precedence order) or you will burn context on unused schemas. Caveat: Kie.ai is an aggregator/reseller path for models from many vendors; rights, availability and ToS safety depend on the underlying provider, and Midjourney/Suno-style access should be treated as gray until verified for your use case.

Also in Model Hubs

fal MCP Serverfal.ai Official

One hosted endpoint to search, run and chain 1,000+ image, video and audio models.

text-to-video · image-to-videoremote · credits

Popcraft MCPPopcraft Official

Hosted OAuth MCP for 20+ video, image, audio and 3D models, with a credit preview on every call.

text-to-video · image-to-videoremote · credits

EmpirioLabs AI MCPEmpirioLabs.ai LLC Official

Remote hub over 180+ hosted models: video, image, speech, 3D, plus GPUs by the hour.

text-to-video · image-to-videoremote · credits

RunAPI MCP ServerRunAPI Official

Browse, price, and run 130+ image, video, audio and voice models through one MCP.

model-search · pricingremote · credits

Imagine MCPImagineArt Official

One OAuth endpoint onto ImagineArt's video, image and music models — 96 tools, no keys.

text-to-video · image-to-videoremote · credits

PiAPI MCP ServerPiAPI

Midjourney in an agent — plus Kling, Luma, Hailuo, Hunyuan, Wan, Suno via PiAPI.

text-to-image · midjourneycredits

Used in

Model Shootout · Compare video and image models before spending the budget

Pairs well with

AKOOL MCPAKOOL Official

Hosted MCP for AKOOL face swap, character swap, image-to-video, TTS and avatar lookup.

face-swap · character-swapremote · credits

Apatero Studio MCPApatero Official

Studio MCP for talking avatars, locked recurring characters, and batched output.

talking-head · avatar-videoremote · credits

bitHuman MCPbitHuman Official

Official bitHuman CLI server: build avatar agents from a face image and render WAV-driven talking video locally.

avatar-video · lip-syncremote · credits