Model Context Protocol

Use Lychhee from Claude

Connect Claude to Lychhee's generation engine via MCP. Generate images, videos, and audio directly from any conversation — billed to your Lychhee credits.

MCP Server URL

https://crexel-backend.onrender.com/mcp

No local install required — paste this URL into your MCP client.

Getting started

Connect in three steps.

1

Create an API key

Go to Settings → Developer and create a key with generate scope. Copy the ck_live_… value — shown once.

Create API key →
2

Connect your client

claude mcp add --transport http lychhee https://crexel-backend.onrender.com/mcp --header "Authorization: Bearer ck_live_…"
3

Ask Claude to generate

Once connected, try: "Generate a cinematic image of a cyberpunk city at night" or "What video models are available?"

Tool reference

8 tools available. Click any tool to expand its parameters and examples.

Example workflows

Real conversations showing how Claude uses Lychhee tools.

Generate a cinematic image

Generate a cinematic image of a samurai standing in a bamboo forest at dusk, warm amber tones
Calling cinema_prompt…cinema_prompt({ subject: "samurai in bamboo forest", palette: "warm amber tones", genre: "cinematic" })
{ "prompt": "samurai in bamboo forest, cinematic cinematography, warm amber tones color palette, cinematic lighting, high dynamic range, ultra-detailed, 8K" }
Calling generate_image…generate_image({ prompt: "…", width: 1280, height: 720 })
{ "status": "completed", "resultUrl": "https://cdn.lychhee.ai/gen/…/image.jpg" }

Create a short video clip

Make a 5-second video of ocean waves at sunset, 16:9 aspect ratio
Calling list_models…list_models({ category: "video" })
{ "models": [{ "id": "fal-ai/kling-video/v2.1/pro/text-to-video", "creditsPerUse": 15 }, …] }
Calling generate_video…generate_video({ prompt: "ocean waves crashing at golden sunset, cinematic", model: "fal-ai/kling-video/v2.1/pro/text-to-video", duration: "5", aspectRatio: "16:9" })
{ "status": "completed", "resultUrl": "https://cdn.lychhee.ai/gen/…/video.mp4" }

Generate a voiceover

Generate a professional voiceover: "Welcome to Lychhee, your AI creative studio"
Calling list_models…list_models({ category: "voice" })
{ "models": [{ "id": "fal-ai/playai-tts", "creditsPerUse": 2 }, …] }
Calling generate_audio…generate_audio({ type: "voice", model: "fal-ai/playai-tts", text: "Welcome to Lychhee, your AI creative studio" })
{ "status": "completed", "resultUrl": "https://cdn.lychhee.ai/gen/…/audio.mp3" }

REST API

For non-MCP integrations. All endpoints are under https://crexel-backend.onrender.com/v1.

Authenticate with Authorization: Bearer ck_live_… or X-API-Key: ck_live_…. Mint a key in Settings → Developer.

MethodPathDescriptionCredits
POST/imageGenerate an image (async, returns jobId)yes
POST/videoGenerate a video (async, returns jobId)yes
POST/audioGenerate voice or music (async, returns jobId)yes
POST/upscale-imageUpscale an image 2× or 4× (synchronous)yes
POST/upscale-videoUpscale a video to 1080p/4K (async)yes
POST/chatAI assistant for creative guidance (synchronous)yes
GET/modelsList all available models with pricingfree
GET/jobs/:idGet status and result URL for a jobfree
GET/openapi.jsonOpenAPI 3.1 machine-readable specfree

Example: generate an image

curl -X POST https://crexel-backend.onrender.com/v1/image \
  -H "Authorization: Bearer ck_live_…" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "A cinematic sunset over the ocean",
    "width": 1280,
    "height": 720
  }'

# Response (202 Accepted):
# { "jobId": "abc-123", "status": "queued", "creditsUsed": 5 }

# Poll for result:
curl https://crexel-backend.onrender.com/v1/jobs/abc-123 \
  -H "Authorization: Bearer ck_live_…"

See the live OpenAPI spec →

FAQ