Content
# Higgsfield MCP API
[](https://nodejs.org/)
[](https://github.com/Noah3521/higgsfield-mcp-api/actions/workflows/ci.yml)
[](./LICENSE)
[](./package.json)
Tiny local REST API for the official Higgsfield MCP endpoint.
English | [한국어](./README.ko.md)
> Unofficial community project. Not affiliated with or endorsed by Higgsfield.
## Why this exists
Higgsfield MCP is useful, but many apps and automation tools want a simple HTTP interface. This project wraps the official MCP tools as a local REST API so you can call Higgsfield from `curl`, Python, Node, n8n, Raycast, local dashboards, or your own scripts.
## Highlights
- Official MCP transport under the hood — no browser automation.
- Simple local REST routes for balance, models, jobs, and image generation.
- Device auth helper with redacted token output.
- Optional `.env` token storage.
- Optional local API key for REST routes.
- Zero npm dependencies.
- English and Korean documentation.
## Requirements
- Node.js 20+
- A Higgsfield account with device auth approval
## Quick start
```powershell
git clone https://github.com/Noah3521/higgsfield-mcp-api.git
cd higgsfield-mcp-api
npm run check
node .\higgsfield-mcp-api.mjs auth-start
```
Open the printed `verification_uri` in your browser and approve the device flow.
Save the access token to `.env` without printing it:
```powershell
node .\higgsfield-mcp-api.mjs auth-save --device-code "PASTE_DEVICE_CODE"
```
Start the API:
```powershell
npm run serve
```
Default base URL:
```txt
http://127.0.0.1:8787
```
## Generate an image
```powershell
curl -X POST http://127.0.0.1:8787/generate/image `
-H "Content-Type: application/json" `
-d '{
"model": "nano_banana_2",
"prompt": "A small red cube on a clean white background",
"count": 1,
"ratio": "1:1",
"params": { "resolution": "1k" }
}'
```
Example response:
```json
{
"results": [
{
"id": "43afc8e6-5fe7-4590-9001-009c8055cfea",
"status": "completed",
"model": "nano_banana_2",
"rawUrl": "https://...png",
"minUrl": "https://...webp"
}
]
}
```
## REST endpoints
Full reference: [docs/API.md](./docs/API.md) / [한국어 API 문서](./docs/API.ko.md)
```txt
GET /health
POST /auth/start
POST /auth/poll
GET /balance
GET /models
GET /models/:id
GET /jobs/:jobId?sync=true
POST /generate/image
GET /tools
POST /tools/:name
```
## CLI examples
```powershell
node .\higgsfield-mcp-api.mjs balance
node .\higgsfield-mcp-api.mjs models --type image --limit 100
node .\higgsfield-mcp-api.mjs generate-image --model nano_banana_2 --prompt "red cube" --ratio 1:1 --resolution 1k
```
## Observed concurrency behavior
Observed on an Ultimate annual account:
```txt
Rate limit reached: max 8 concurrent job(s) on ultimate (annual) plan.
```
The server maps this MCP response to HTTP `429`.
## Security notes
- Bind to `127.0.0.1` by default.
- Never commit `.env`, tokens, device codes, cookies, session values, or HAR files.
- Set `HIGGSFIELD_MCP_API_KEY` if another local process can access the server.
- MCP does not expose the web-only `use_unlim` flag, so image generation may consume credits.
## Roadmap
- More model-specific examples.
- Dockerfile and compose example.
- Typed client snippets for Python and TypeScript.
- Optional local request queue for friendlier concurrency behavior.
## Contributing
Issues and pull requests are welcome. See [CONTRIBUTING.md](./CONTRIBUTING.md).
If this helped you, please consider starring the repo — it makes the project easier for other builders to discover.
Connection Info
You Might Also Like
markitdown
Python tool for converting files and office documents to Markdown.
OpenAI Whisper
OpenAI Whisper MCP Server - 基于本地 Whisper CLI 的离线语音识别与翻译,无需 API Key,支持...
oh-my-opencode
Background agents · Curated agents like oracle, librarians, frontend...
claude-flow
Claude-Flow v2.7.0 is an enterprise AI orchestration platform.
ai-engineering-from-scratch
Learn it. Build it. Ship it for others. The most comprehensive open-source...
chatbox
User-friendly Desktop Client App for AI Models/LLMs (GPT, Claude, Gemini, Ollama...)