Let your agent make thumbnails.
Connect an assistant through MCP, drop in the Agent Skill, or let a wallet-enabled agent pay per call with x402. Machine-readable summary at /llms.txt.
# Claude Code
claude mcp add --transport http thumbnailcreator https://thumbnailcreator.ai/api/mcp \
--header "Authorization: Bearer tc_your_key"
# Cursor, Claude Desktop, ChatGPT and other MCP clients
{
"mcpServers": {
"thumbnailcreator": {
"url": "https://thumbnailcreator.ai/api/mcp",
"headers": { "Authorization": "Bearer tc_your_key" }
}
}
}
# Grok (xAI Responses API, remote MCP tool)
{ "type": "mcp", "server_url": "https://thumbnailcreator.ai/api/mcp",
"server_label": "thumbnailcreator", "authorization": "tc_your_key" }MCP server
Streamable HTTP at https://thumbnailcreator.ai/api/mcp. Sign in, open the account menu and create a key under Agent keys, then send it as Authorization: Bearer tc_…. Usage bills to your account at the per-image prices below, paid from your credit balance.
| list_models | Models with the price per image. |
| generate_thumbnails | prompt, models?, n?, format?. Up to 8 images; waits and returns image URLs plus the amount charged. |
| get_generation | Status and fresh URLs for a generation id. |
| account_usage | Credits left, credits bought and in-flight spend. |
Claude.ai and other OAuth clients
Apps that connect with sign-in instead of a key (Claude.ai web, desktop and mobile, ChatGPT connectors) use https://thumbnailcreator.ai/api/connector/mcp. In Claude, open Settings › Connectors › Add custom connector, paste the URL and sign in. Same tools and prices. Each connection appears under Agent keys, where you can revoke it.
Agent Skill
SKILL.md teaches Claude and other skill-aware agents when to make a thumbnail, which models to pick and how to prompt them. It uses the MCP server when connected and x402 otherwise.
mkdir -p ~/.claude/skills/thumbnailcreator
curl -o ~/.claude/skills/thumbnailcreator/SKILL.md https://thumbnailcreator.ai/skill.mdPay per call, no account
POST https://thumbnailcreator.ai/api/x402/generations with pricing inputs in the query string (models, n, format) and { "prompt": "…" } in the body. The server answers 402 with the exact USDC quote; an x402 client signs it and retries. The call waits for the images and returns their URLs. Payment settles only if every image succeeds; a partial failure or a timeout is free and still returns the finished images. Fetch results again at GET /api/x402/generations/:id.
POST https://thumbnailcreator.ai/api/x402/generations?models=gpt-image-2,grok-imagine&n=1&format=16:9
{ "prompt": "Chef tasting a giant burger, bold text \"$500 BURGER\"" }
→ 402 PAYMENT-REQUIRED: { amount, asset: USDC, network, payTo }
→ 200 { "status": "succeeded", "images": [{ "model": "gpt-image-2", "url": "https://…" }, …] }Models and prices
Formats: 16:9 (1920×1080), 1:1 (1536×1536), 9:16 (1080×1920), 4:5 (1200×1500). Up to 4 models × 4 images, 8 images per request. Failed images are free.
| gpt-image-2 | GPT Image 2 | $0.167 |
| gpt-image-2.5-flare | GPT Image 2.5 Flare | $0.062 |
| gpt-image-2.5-sunburst | GPT Image 2.5 Sunburst | $0.167 |
| nano-banana-pro | Nano Banana Pro | $0.195 |
| nano-banana-2.1 | Nano Banana 2.1 | $0.066 |
| flux-2-max | FLUX.2 Max | $0.130 |
| seedream-5-lite | Seedream 5 Lite | $0.046 |
| ideogram-v4 | Ideogram 4 | $0.130 |
| grok-imagine | Grok Imagine | $0.026 |