Quickstart
Getting started
Connect any MCP-compatible client to the WM Studio MCP and run your first generation in minutes.
Prerequisites
- A WM Studio account (OAuth flow signs you in via Supabase).
- An MCP client — Claude.ai, Claude Code, Cursor, or the MCP Inspector.
- The public MCP URL, e.g.
https://<your-fly-app>.fly.dev/mcp/.
Connect Claude
- Open Claude.ai → Settings → Connectors.
- Add custom connector with your
MCP_BASE_URL+/mcp/. - Complete OAuth in the browser (Supabase IdP).
Connect Cursor / Claude Code
mcp.json
{
"mcpServers": {
"wmstudio": {
"url": "https://<your-fly-app>.fly.dev/mcp/",
"auth": "oauth"
}
}
}Local development
Run the server locally with Redis for OAuth state and job tracking.
python3 -m venv .venv && source .venv/bin/activate
pip install -e '.[dev]'
cp .env.example .env
docker compose up redis -d
PYTHONPATH=. uvicorn src.server:root_app --reload --host 0.0.0.0 --port 8080Your first tool call
The fastest path: a single studio_generate_image call. It returns a hosted image URL on the WM Studio CDN that you can pipe into any other tool.
json
{
"name": "studio_generate_image",
"arguments": {
"prompt": "minimalist product photo of a black ceramic mug, soft daylight",
"model": "fal-ai/flux-2-pro",
"aspect_ratio": "4:3"
}
}Want to refine? Feed the returned URL into studio_upscale_image for 2×/4× detail, or into studio_camera_angles for a 9-shot exploration.
Next steps
- Browse the tool reference.
- Chain calls in workflows.
- Map common failures in errors.