Kie Mcp

@dainami/kie-mcp

> Let Claude generate images, videos, and music with the latest AI models — Seedance, Kling, GPT Image 2, Nano Banana, Suno V4, and anything new KIE ships — without leaving the chat.

A tiny MCP server that lets Claude (Desktop, Code, or any MCP client) call any model on KIE.ai on your behalf. Make a 9:16 video with Seedance 2.0 Pro. Edit a product photo with Nano Banana 2. Render a 5-panel storyboard with GPT Image 2. Compose a soundtrack with Suno V4.

It's intentionally dumb on purpose: 5 generic tools instead of one tool per model. Claude reads the JSON shape from docs.kie.ai when it needs to, and constructs the payload itself. When KIE ships a new model tomorrow, this package doesn't have to update — Claude just points at the new docs.

---

Which install do I need?

Pick the row that matches how you run Claude:

• Claude Code (terminal) → use Quick start: Claude Code below — one command, the fastest option.
• Regular Claude Desktop chat → use Path A (drag-and-drop .mcpb, 2 minutes, no terminal).
• Claude Desktop co-work → use Path B (download source, build once, edit a config file). Co-work ignores .mcpb files, so you set it up by hand.

If you use both Desktop modes, do Path B — it works for both.

---

Quick start: Claude Code (fastest)

One command — no .mcpb, no JSON editing.

1. Get a KIE.ai API key at kie.ai → Dashboard → API Keys.
2. Clone + build (once, somewhere permanent):

``bash git clone https://github.com/mrdainami/kie-mcp.git ~/mcp/kie-mcp cd ~/mcp/kie-mcp && npm install && npm run build ``

1. Register it (--scope user makes it available in every project; drop it to add only to the current project):

``bash claude mcp add --scope user kie --env KIE_API_KEY=YOUR_KEY -- node ~/mcp/kie-mcp/dist/index.js ``

1. Verify: claude mcp list should show kie ✓ Connected.

Your key lives in Claude Code's own config, never in this repo. To update later: git pull && npm run build in the folder, then restart Claude Code.

---

Path A — Regular Claude Desktop (drag-and-drop)

1. Get a KIE.ai API key at kie.ai → Dashboard → API Keys. Copy it somewhere — you'll paste it in step 4.
2. Download the latest .mcpb from the releases page.
3. Drag that file onto the Claude Desktop window (or open Settings → Extensions and pick it).
4. When Claude pops up a box asking for your KIE API key, paste it.
5. Quit Claude Desktop fully and reopen it.

Done. To check it's connected, open a chat → click the + button → Connectors — you should see "KIE.ai" listed with its tools.

> Why .mcpb? It's Anthropic's drag-and-drop install format for local MCP servers. No JSON editing, no Node.js install required (Claude Desktop bundles its own Node runtime). Read more →

---

Path B — Claude co-work (manual install)

Co-work doesn't load .mcpb files. You have to install Node.js, download this repo, build it once, then point Claude at the built file via the config JSON. One-time setup, ~10 minutes.

1. Get your API key

Sign up at kie.ai → Dashboard → API Keys → copy the key. Keep it handy.

2. Install Node.js

If you don't have it: go to nodejs.org → click the big LTS download button → run the installer → click Next a few times. Done. This adds Node to your system so the MCP can run.

3. Download this repo and build it

You only do this once. The folder you create here is permanent — Claude will look at it forever, so put it somewhere you won't move or delete (~/Documents/kie-mcp is a good spot).

• Go to github.com/mrdainami/kie-mcp
• Click the green Code button → Download ZIP
• Unzip it. Rename the folder from kie-mcp-main to kie-mcp if you like, and put it in ~/Documents/.
• Open Terminal (Spotlight → type "Terminal") and run these three lines, one at a time:

``bash cd ~/Documents/kie-mcp npm install npm run build ``

After npm run build finishes, you can close Terminal and never open it again.

4. Tell Claude where to find it

• In Finder, press Cmd+Shift+G and paste this path, then Enter:

`` ~/Library/Application Support/Claude/ ``

• Open claude_desktop_config.json in TextEdit (or VS Code, any text editor).
• If the file is empty, paste the whole block below. If it already has stuff, just add the "kie": { ... } block inside the existing "mcpServers" object:

``json { "mcpServers": { "kie": { "command": "node", "args": [ "/Users/YOUR_USERNAME/Documents/kie-mcp/dist/index.js" ], "env": { "KIE_API_KEY": "paste-your-kie-key-here" } } } } ``

• Replace YOUR_USERNAME with your Mac username (Terminal whoami tells you if you don't know).
• Replace paste-your-kie-key-here with the key from step 1.
• Save the file. Quit Claude Desktop fully and reopen it.

On Windows, the config file lives at %APPDATA%\Claude\claude_desktop_config.json and the args path will be a Windows-style path like C:\\Users\\You\\Documents\\kie-mcp\\dist\\index.js.

5. Try it

In co-work, ask: "Make me a 9:16 video of a coffee cup using Seedance." Claude will submit the job with kie_post, poll with kie_get, and return the video URL when it's done.

If something goes wrong:

• "command not found: npm" → Node.js isn't installed. Redo step 2.
• "module not found" / "ENOENT dist/index.js" → you skipped npm run build in step 3. Run it again.
• The MCP shows up but every call fails with an auth error → your KIE_API_KEY is wrong or missing. Re-check step 4.

---

What the agent does with these tools

Generation on KIE is asynchronous — submit a job, then poll for the result. Typical flow for "make me an image with GPT Image 2":

1. Agent (optionally) calls kie_fetch_model_docs to confirm the model's current JSON shape.
2. Agent submits with kie_post:

``ts kie_post({ path: "/api/v1/jobs/createTask", body: { model: "gpt-image-2", input: { prompt: "A serene cabin in the mountains, golden hour", aspect_ratio: "9:16", quality: "high" } } }) // → body.data.taskId — save this immediately ``

1. Agent polls with kie_get every ~20–30s until done:

``ts kie_get({ path: "/api/v1/jobs/recordInfo?taskId=..." }) // data.state: waiting | generating | success | fail // on success → data.resultJson.resultUrls[] ``

1. Agent saves the result with kie_download({ url, destPath }).

> Cost is billed on submit, not on poll. Persist the taskId right after submitting, and never resubmit a live task.

---

The 5 tools

| Tool | What it does | |---|---| | kie_post(path, body) | POST to any KIE endpoint — submit a generation task (usually /api/v1/jobs/createTask). | | kie_get(path) | GET from any KIE endpoint — poll task status (usually /api/v1/jobs/recordInfo?taskId=...). | | kie_upload_file(localPath, uploadPath?) | Upload a local file to KIE storage → returns a hosted URL (~3-day TTL) for use as an @Image/@Video reference. | | kie_download(url, destPath) | Download a result URL to local disk (creates parent folders). | | kie_fetch_model_docs(path \| url, force?) | Fetch a model's live docs from docs.kie.ai (cached ~3 days) so the agent knows the exact payload shape. |

Different model families use slightly different envelopes (Veo, Suno, Flux-Kontext) — the agent reads the model docs and adjusts. The MCP stays generic, so new KIE models work without updating this package.

---

Environment variables

| Variable | Default | Required | |---|---|---| | KIE_API_KEY | — | Yes | | KIE_BASE_URL | https://api.kie.ai | No (override for testing / proxy) |

---

Develop locally

git clone https://github.com/mrdainami/kie-mcp
cd kie-mcp
npm install
npm run build

Point Claude at your local copy:

{
  "mcpServers": {
    "kie": {
      "command": "node",
      "args": ["/absolute/path/to/kie-mcp/dist/index.js"],
      "env": { "KIE_API_KEY": "kie-..." }
    }
  }
}

For dev with auto-reload:

KIE_API_KEY=kie-... npm run dev

---

Why it's tiny

Most MCP wrappers around AI APIs hard-code one tool per model: generate_seedance(), generate_gpt_image_2(), generate_suno(). That means:

• You have to update the wrapper every time the provider ships a new model
• The wrapper's mental model of each model's JSON shape drifts from reality
• The wrapper becomes the bottleneck

This package goes the opposite direction: three generic tools that work for everything KIE has, with KIE's docs as the source of truth for shapes. Claude is smart enough to read docs and construct payloads. The MCP just makes the HTTP call.

When KIE ships a new model tomorrow, no PR here. Just point your agent's knowledge files at the new docs URL.

---

License

MIT — see [LICENSE](./LICENSE).

By mrdainami. PRs and issues welcome.

Source & license

This open-source MCP server is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: mrdainami
• Source: mrdainami/kie-mcp
• License: MIT

Install and usage instructions live in the source repository linked above.