Thumbnail Creator

Generate professional YouTube thumbnails using AI. This skill connects to the Thumbnail Creator API to create thumbnails, analyze videos, manage creative assets, and generate SEO content.

Setup

Before using this skill, ensure you have:

1. A Thumbnail Creator account at https://app.thumbcraft.io
2. An API key from Settings → Developers → API Keys

MCP Configuration (Recommended)

Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json on macOS):

{
  "mcpServers": {
    "thumbnail-creator": {
      "url": "https://app.thumbcraft.io/api/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

REST API

Base URL: https://app.thumbcraft.io/api/v1

All requests require the header:

Authorization: Bearer YOUR_API_KEY

Available Tools

Thumbnail Generation

generate_thumbnail

Create a thumbnail from a text description.

Parameters:

• prompt (required): Detailed description of the thumbnail
• youtube_url (optional): YouTube URL for context
• style_id (optional): Apply a saved style
• face_ids (optional): Array of face IDs to include
• brand_id (optional): Brand ID for colors/logo
• aspect_ratio (optional): "16:9" (default) or "9:16"

Best practices for prompts:

• Be specific about composition, colors, and emotions
• Mention text overlays explicitly if needed
• Describe the main subject's expression and pose
• Include background details

Example:

Generate a thumbnail with a surprised man looking at a giant golden trophy.
Bold yellow text says "I WON". Dark gradient background with sparkles.
Photorealistic style, dramatic lighting.

analyze_video

Analyze a YouTube video and get 5 thumbnail concepts with prompts.

Parameters:

• youtube_url (required): Full YouTube URL

Returns: Video summary and 5 thumbnail concepts with:

• Title and description
• Ready-to-use generation prompt
• Emotional hook explanation

Asset Management

list_faces

Get uploaded face models for personalized thumbnails.

Parameters:

• active_only (optional): Only return trained faces

create_face

Create a new face from an image URL. Requires training before use.

Parameters:

• name (required): Display name
• image_url (required): Clear face photo URL

list_styles

Get saved thumbnail styles to apply to generations.

list_brands

Get brand assets (logos, colors) for consistent branding.

list_thumbnails

Get recent generated thumbnails.

Parameters:

• limit (optional): Number to return (default: 10)

SEO Tools

generate_seo_titles

Generate 10 SEO-optimized video titles from a YouTube URL.

Parameters:

• youtube_url (required): YouTube video URL

Returns: 10 titles with style labels:

• Search-Driven (keyword-focused)
• Curiosity/Viral (click-driving)
• Authority/Expertise (trust-building)
• Emotional/Story (connection)
• Contrarian/Controversial (attention-grabbing)

generate_seo_description

Generate a complete video description.

Parameters:

• youtube_url (required): YouTube video URL

Returns structured description with:

• Hook (2 engaging sentences)
• About This Video (150-200 word summary)
• Key Takeaways (3-5 bullet points)
• Timestamps/Chapters (6-10 sections)
• Notable Quotes (1-2)
• SEO Hashtags (3-5)

get_credits

Check remaining credit balance.

Workflows

Generate Thumbnail from YouTube Video

1. Call analyze_video with the YouTube URL
2. Present the 5 concepts to the user
3. Let user pick a concept or provide feedback
4. Call generate_thumbnail with the chosen concept's prompt
5. Optionally apply a style or include faces

Create Thumbnails with Consistent Style

1. Call list_styles to show available styles
2. Let user pick a style or describe a new one
3. Call generate_thumbnail with style_id parameter

Full YouTube Optimization

1. Call analyze_video for thumbnail concepts
2. Generate thumbnail with generate_thumbnail
3. Call generate_seo_titles for title options
4. Call generate_seo_description for the description
5. Present complete package: thumbnail + title + description

Tips for Best Results

Thumbnail Prompts

• Be specific: "A woman with wide eyes and open mouth" > "surprised person"
• Describe lighting: "dramatic side lighting", "soft natural light"
• Mention style: "photorealistic", "3D rendered", "illustrated"
• Include text carefully: Specify exact text, font style, and placement
• Consider contrast: High contrast between subject and background

Using Faces

• Faces must be trained before use (check trainingStatus)
• Use face_ids array to include multiple people
• The AI will match the face to the described expression

Using Styles

• Styles capture visual aesthetics from reference images
• Great for maintaining channel consistency
• Combine with custom prompts for unique results

Error Handling

Common errors:

• 401 Unauthorized: Check API key
• 402 Payment Required: Out of credits
• 429 Too Many Requests: Rate limited, wait and retry
• Invalid YouTube URL: Ensure full URL format

Credits

Operations that consume credits:

• generate_thumbnail: 1 credit
• All other operations: Free

Check balance with get_credits before generating.