This server provides AI image and video generation capabilities via the Gwanggo API, supporting 35+ models (Seedream, Sora, Veo, Kling, GPT Image, and more) through a single API key.
list_models: Browse 35+ image and video generation models, optionally filtering by type, with credit costs and per-model options (aspect ratios, resolutions, durations).
generate_image: Create images from text prompts or edit existing images using a reference URL; supports model selection, quality tiers, and aspect ratio settings.
generate_video: Create videos from text prompts or reference images; supports model selection, duration, resolution, aspect ratio, and optional audio track generation; waits up to 10 minutes for completion.
get_task: Retrieve the status and result of a previously submitted image/video generation job by its ID.
get_credits: View the connected Gwanggo account email and remaining credit balance.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@gwanggo-mcp generate an image of a sunset over mountains"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
MCP server + CLI for the Gwanggo API — Seedream, Sora, Veo, Kling, GPT Image and 30+ more models behind one key.
Works with Claude Code · Claude Desktop · Cursor · Codex — anything that speaks MCP.
You: "우리 신제품 세럼 광고 이미지 만들어줘, 화장품 무드로"
Agent: [calls generate_image with seedream-5] → https://…/result.png1. Connect your account (opens a browser approval — no password pasting):
npx gwanggo-mcp login
# or, after install:
gwanggo auth loginNew accounts get free credits: sign up at gwanggo.ai. You can also skip
loginand setGWANGGO_API_KEY(create one at Dashboard → API keys).
2. Add to your agent:
Claude Code
claude mcp add gwanggo -- npx -y gwanggo-mcpClaude Desktop / Cursor (claude_desktop_config.json / .cursor/mcp.json)
{
"mcpServers": {
"gwanggo": {
"command": "npx",
"args": ["-y", "gwanggo-mcp"]
}
}
}Codex CLI (~/.codex/config.toml)
[mcp_servers.gwanggo]
command = "npx"
args = ["-y", "gwanggo-mcp"]To pass the key explicitly instead of login, add "env": { "GWANGGO_API_KEY": "gwk_..." }.
Related MCP server: Corent MCP Server
Tool | What it does |
| 35+ image/video models with credit costs and per-model options |
| Text→image / image edit. Waits for completion, returns the URL |
| Text→video / image→video (Seedance 2.0, Kling 3.0, Veo 3.1, Sora 2 …) |
| Check a previous generation by id |
| Remaining credit balance |
Failed generations are auto-refunded server-side.
The same binary doubles as a CLI:
gwanggo auth login
gwanggo models
gwanggo generate image "neon alley cat, cinematic" --model gpt-image-2 --quality high
gwanggo generate video "waves crashing at dawn" --model seedance-2.0 --resolution 720p --duration 5 --generate-audio
gwanggo me # account and credits
gwanggo task <id> # check statusThe gwanggo commands reuse the account saved by gwanggo auth login; no API-key
environment variable is required. GWANGGO_API_KEY remains available for CI and
other non-interactive environments.
Thin client over the public Gwanggo /v1 REST API — the same API that powers the open-source Gwanggo Studio. Your key is stored at ~/.config/gwanggo/config.json (0600) and sent only as a Bearer token. A browser-login key takes precedence over GWANGGO_API_KEY, so a stale environment key cannot override a newly approved login. In CI, where no login file exists, GWANGGO_API_KEY is used.
Generate an image with a Gwanggo model. Spends credits. Waits for completion and returns the image URL. Use list_models first to choose a model slug and see its options.
Inputs: model (required), prompt (required), quality, image_url, +1 more.
Generate a video with a Gwanggo model. Spends credits (often per-second — check list_models). Waits up to 10 minutes and returns the video URL.
Inputs: model (required), prompt (required), duration, image_url, +3 more.
Get the connected Gwanggo account email and remaining credit balance.
Inputs: no named parameters.
Check the status/result of a previous generation by its id.
Inputs: id (required).
List available AI image/video generation models with credit costs and per-model options (aspect ratios, resolutions, durations). Call this first to pick a model slug.
Inputs: type.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 5 tools
Each tool serves a distinct function: generating images, generating videos, checking credits, checking task status, and listing models. No overlaps or ambiguity.
All tool names follow a consistent verb_noun pattern (generate_image, generate_video, get_credits, get_task, list_models), making predictions easy.
With 5 tools, the set covers the core workflow for an AI media generation service without being too sparse or excessive.
The surface covers model discovery, generation for both image and video, credit tracking, and task status. Missing a cancel/delete tool is a minor gap but not critical.
Generate images and videos with 200+ AI models (GPT-Image-2, Seedance, Veo, Kling) via one API.
Generate images, video, music, voice and 3D through one API. 30 tools, 200+ models.
Generate images and videos (GPT Image, Seedance, Kling, Wan, Qwen) with one API key.
Best Image and video generation: 20+ models (Kling, Seedance, Veo, NB, FLUX.2), OAuth, pay-per-use.
