Gamma claude skill

Gamma.app Content Optimization

Transform raw content into professional Gamma-optimized output using structure-first methodology and tier-appropriate settings.

Core Capabilities

• Convert unstructured content into Gamma-ready format with --- delimiters
• Apply tier-specific constraints (Free: 10 cards, Plus: 20, Pro: 60, Ultra: 75)
• Generate SPLICE-formatted image prompts for AI image generation
• Provide comprehensive Gamma UI settings and API parameter mappings
• Create post-generation refinement plans using Gamma Agent commands
• Handle multiple formats: presentations, documents, webpages, social posts

Three-Stage Workflow

Execute this process automatically when users provide content:

Stage 1: Analysis

• Assess content volume, structure, complexity, and format goal
• Apply constraint analysis based on user's Gamma tier
• Verify current Gamma capabilities (features evolve frequently)
• Output brief diagnostic summary

Stage 2: Transformation

• Restructure content with --- card breaks (structure-first methodology)
• Condense if exceeding tier limits
• Map optimal Gamma API parameters
• Develop image generation strategy using SPLICE framework
• Create tone/audience specifications

Stage 3: Dual Output Delivery

Always deliver two artifacts:

Artifact 1: Paste-Ready Gamma Input

• Title format: [Project Name] - Gamma Input [Format]
• Structured content with --- delimiters between cards
• Clear H1/H2 headings for each section
• Organized bullet points and sub-points
• Ready to paste directly into Gamma with no cleanup

Artifact 2: Optimization Guide

• Title format: [Project Name] - Gamma Optimization Guide
• Section I: Recommended Gamma settings (all parameters)
• Section II: Image generation strategy with SPLICE prompts
• Section III: Gamma Agent usage plan (post-generation commands)
• Section IV: Known limitations & workarounds
• Section V: Post-generation checklist

Chat Response:

• Transformation summary (2-3 sentences)
• Critical recommendations (3-5 bullets)
• Next steps (immediate + follow-up actions)

Structure-First Formatting

Apply this pattern for all Gamma input:

# [Primary Topic/Title]
[Optional: Brief context or hook]

---

# [First Major Section]
- Key point 1
- Key point 2
- Key point 3

---

# [Second Major Section]
[Content structured for this card]

---

Key Principles:

• Use --- to force card breaks (inputTextBreaks mode)
• Create clear, descriptive H1/H2 headings
• Balance content distribution across cards
• Front-load most important information
• No inline citations or bracketed references

Gamma API Parameters

Always provide these parameters (even for UI users). Based on official API v1.0 specification:

Core Parameters (Required):

• inputText: The restructured content (string, max ~100,000 tokens)
• textMode: "generate" | "condense" | "preserve" (required)

Core Parameters (Optional):

• format: "presentation" | "document" | "webpage" | "social" (default: "presentation")
• numCards: Integer, recommended count (respects tier limits: Free=10, Plus=20, Pro=60, Ultra=75)
• cardSplit: "inputTextBreaks" | "auto" (use "inputTextBreaks" when content has --- delimiters)
• themeId: String, specific theme ID (get via List Themes API)
• folderIds: Array of strings, save to specific folders
• exportAs: "pdf" | "pptx" | null (export immediately after generation)
• additionalInstructions: String, extra guidance (max 2000 chars)

Text Options (Nested Object):

"textOptions": {
  "amount": "minimal" | "concise" | "detailed" | "extensive",
  "tone": "professional, inspiring",  // free-form string
  "audience": "C-suite executives",    // free-form string
  "language": "en"                      // ISO 639-1 code (60+ languages)
}

Image Options (Nested Object):

"imageOptions": {
  "source": "aiGenerated" | "unsplash" | "noImages",
  "model": "imagen-4-pro",  // model ID string (see Image Models section)
  "style": "photorealistic" // free-form style keywords
}

Card Options (Nested Object with Header/Footer):

"cardOptions": {
  "dimensions": "16:9" | "4:3" | "fluid",
  "headerFooter": {
    "topRight": {"type": "image", "source": "themeLogo", "size": "sm"},
    "bottomRight": {"type": "cardNumber"},
    "hideFromFirstCard": true,
    "hideFromLastCard": false
  }
}

Sharing Options (Nested Object):

"sharingOptions": {
  "workspaceAccess": "view" | "comment" | "edit" | "noAccess",
  "externalAccess": "view" | "noAccess",
  "emailOptions": {
    "recipients": ["[email protected]"],
    "access": "view" | "comment" | "edit"
  }
}

SPLICE Image Prompting Framework

For AI-generated images, structure prompts with:

S - Style & Medium Examples: "watercolor painting", "3D render", "photorealistic", "flat icon style"

P - Perspective & Composition Examples: "close-up portrait", "bird's-eye view", "wide-angle centered"

L - Lighting & Atmosphere Examples: "soft morning light", "dramatic shadows", "neon glow"

I - Identity of Subject Examples: "a knight in armor", "modern office building"

C - Cultural/Contextual Details Examples: "Victorian era furniture", "snow-covered ground"

E - Emotion & Energy Examples: "joyful exuberance", "serene and peaceful"

Example Complete Prompt: "A vintage oil painting (Style) of a cat close-up (Perspective), candlelit in a dark room (Lighting), the cat wearing a small top hat (Identity), sitting on an old piano with sheet music (Context), looking mysterious and regal (Emotion)."

Tier-Specific Constraints

Free Tier

• 10 cards per generation max
• Basic image models only (Flux Fast 1.1, Imagen 3 Fast)
• Limited analytics, Gamma branding on exports, no custom themes
• Strategy: Keep content concise, focus on core message

Plus Tier ($10/month)

• 20 cards per generation max
• Advanced image models (Flux Pro, Imagen 3)
• Basic analytics, can hide Gamma branding
• Strategy: Good for personal/small projects

Pro Tier (~$25/month)

• 60 cards per generation max
• Premium image models (DALL·E 3, Imagen 4, Leonardo Phoenix)
• Full analytics, custom themes, API access
• Strategy: Recommended for professional use, brand consistency

Ultra Tier

• 75 cards per generation max
• Ultra-exclusive models (Flux Ultra, GPT-Image-1 High)
• 20,000 credits/month vs 4,000 (Pro), video generation
• Strategy: For power users with high volume

Image Model Selection by Tier

Based on official API documentation. Use exact model IDs in imageOptions.model:

Free Tier Models (2 credits/image):

• flux-fast-1-1 - Flux Fast 1.1 (fast, decent quality)
• imagen-3-fast - Imagen 3 Fast (Google quick model)

Plus Tier Models (8-10 credits/image):

• flux-pro - Flux Pro (high quality, versatile)
• imagen-3 - Imagen 3 (Google standard)
• ideogram-3-turbo - Ideogram 3 Turbo (good for text in images)

Pro Tier Models (15-33 credits/image):

• imagen-4-pro - Imagen 4 (best overall, Google latest) ⭐ RECOMMENDED
• dall-e-3 - DALL·E 3 (OpenAI, creative)
• leonardo-phoenix - Leonardo Phoenix (artistic, stylized)

Ultra Tier Models (30-120 credits/image):

• flux-ultra - Flux Ultra (absolute best quality)
• imagen-4-ultra - Imagen 4 Ultra (Google premium)
• gpt-image-1-high - GPT-Image-1 High (cutting-edge)

Special Note: Model availability and names should be verified via the Image Model API endpoint if uncertain

Gamma Agent Post-Generation Commands

Recommend specific commands based on content needs:

Common Agent Tasks:

• "Make all headings title case"
• "Add a new card summarizing key takeaways"
• "Search for recent statistics on [topic] and add with citations"
• "Translate this card to [language]"
• "Change tone of this card to be more conversational"
• "Create a timeline diagram on this card"
• "Fix grammar and spelling across all cards"
• "Apply [theme] and ensure consistent formatting"

Agent Capabilities:

• Web search with citations
• Bulk edits across multiple cards
• Content generation in context
• Theme application
• Translation
• Image generation and placement
• Reading external URLs or uploaded content

Text Mode Decision Logic

Generate: Use when input is brief outline/skeleton → Gamma expands it with AI-generated content

• Best for: Topic lists, bullet points, rough notes
• Gamma writes full sentences and paragraphs
• API value: "textMode": "generate"

Condense: Use when content is long → Gamma summarizes/shortens

• Best for: 20+ page documents, dense reports
• Gamma extracts key points and makes concise
• API value: "textMode": "condense"

Preserve: Use when content is optimized → Gamma keeps exact wording

• Best for: Content already polished by Claude
• Gamma uses text as-is, just adds formatting/layout
• MOST COMMON when using this skill
• API value: "textMode": "preserve"

Default recommendation: "preserve" (since this skill optimizes text first)

Theme Recommendations

Title: Dark blue, professional, corporate Stratos: Dark navy, tech-forward, serious Lux: Dark green, elegant, finance-oriented Prism: Light, colorful accents, creative Basic Light: Clean white, minimal, universal (safe default) Aurora: Purple/blue gradient, modern, dynamic

Special Scenarios

Very Short Input (< 500 words):

• May fit single generation easily
• Focus on polish and structure
• Ensure user isn't underselling scope

Very Long Input (> 20,000 words):

• Recommend splitting into multiple Gammas
• Provide priority-based split: "core + supplemental"
• Explain phasing strategy

Technical/Data-Heavy Content:

• Consider chart/table creation in Gamma
• Flag: No live data binding (static snapshots only)
• Provide data visualization strategy

Highly Visual Content:

• Emphasize SPLICE framework application
• Recommend appropriate image models by tier
• Budget for credit costs and regeneration needs

Complete API Request Example

Based on official Gamma API v1.0 specification:

curl --request POST \
  --url https://public-api.gamma.app/v1.0/generations \
  --header 'Content-Type: application/json' \
  --header 'X-API-KEY: <your-api-key>' \
  --data '{
    "inputText": "# Best Hikes in the United States\n\n---\n\n# Yosemite National Park\n- Half Dome trail\n- Mist Trail to Vernal Fall\n\n---\n\n# Grand Canyon\n- Bright Angel Trail\n- South Kaibab Trail",
    "textMode": "preserve",
    "format": "presentation",
    "numCards": 10,
    "cardSplit": "inputTextBreaks",
    "themeId": "<theme-id-from-list-themes-api>",
    "folderIds": ["<folder-id>"],
    "exportAs": "pdf",
    "additionalInstructions": "Make the titles catchy and emphasize natural beauty",
    "textOptions": {
      "amount": "detailed",
      "tone": "professional, inspiring",
      "audience": "outdoors enthusiasts, adventure seekers",
      "language": "en"
    },
    "imageOptions": {
      "source": "aiGenerated",
      "model": "imagen-4-pro",
      "style": "photorealistic"
    },
    "cardOptions": {
      "dimensions": "16:9",
      "headerFooter": {
        "topRight": {
          "type": "image",
          "source": "themeLogo",
          "size": "sm"
        },
        "bottomRight": {
          "type": "cardNumber"
        },
        "hideFromFirstCard": true,
        "hideFromLastCard": false
      }
    },
    "sharingOptions": {
      "workspaceAccess": "view",
      "externalAccess": "noAccess",
      "emailOptions": {
        "recipients": ["[email protected]"],
        "access": "comment"
      }
    }
  }'

Success Response:

{
  "generationId": "abc123xyz"
}

Error Response (400):

{
  "message": "Input validation errors: 1. numCards exceeds tier limit",
  "statusCode": 400
}

Error Response (402 - No Credits):

{
  "message": "Insufficient credits",
  "statusCode": 402
}

Quality Checklist

Before delivering output, verify:

• Content respects card/token limits for user's tier
• Structure uses --- breaks appropriately (if using inputTextBreaks mode)
• All headings are clear and descriptive (H1/H2 format)
• Image strategy is specific with SPLICE framework prompts
• Recommendations are tier-appropriate (model availability, card limits)
• API parameters use correct names from official v1.0 spec
• Nested objects (textOptions, imageOptions, etc.) properly structured
• Image model IDs are exact strings (e.g., "imagen-4-pro" not "Imagen 4 Pro")
• Optimization guide includes actionable Gamma Agent commands
• Known limitations explicitly flagged with workarounds
• Output is paste-ready with no cleanup required
• No inline citation markers or bracketed references
• Export format specified if needed ("pdf" | "pptx" | null)
• Sharing permissions configured if collaboration required

Constraint Communication

• State limits explicitly with numbers: "This is 150 cards of content; Gamma Pro limit is 60 per generation"
• Provide specific recommendations, not vague suggestions
• Offer splitting strategies when content exceeds limits
• Explain trade-offs transparently

Important Notes

Input Limits:

• ~100,000 tokens (~75,000 words) max input
• Large content may require splitting

Export Limitations:

• Interactive elements (toggles, embeds) flatten in PDF/PPTX
• Custom fonts need local installation for PPTX viewing
• Gradient text may become single color in PowerPoint export

Version Control:

• Gamma evolves frequently
• Always verify current capabilities if uncertain
• If features seem outdated, note this to user

Gamma MCP Integration

When working with Gamma MCP (Model Context Protocol) connector:

Available via MCP:

• Gamma MCP server provides three core tools to AI assistants:
  1. Generate content (presentations, documents, webpages, social posts)
  2. Create from template (use existing Gamma as starting point)
  3. List available themes and folders

Setup for Claude Desktop/Web:

1. Go to Settings → Connectors → Browse Connectors
2. Search for "Gamma" and click Connect
3. Authorize access to your Gamma account
4. Select workspace (if applicable)

MCP Best Practices:

• Be specific about structure and content in requests
• Specify format, card count, theme preferences upfront
• Provide SPLICE-formatted image descriptions for AI images
• MCP mirrors Generate API capabilities (all parameters available)
• If MCP returns warnings, Claude will suggest corrected values

MCP vs Direct API:

• Use MCP: When building AI tool integrations (Claude, other assistants)
• Use API: When building custom apps requiring programmatic control
• Use Automation Platforms: Zapier/Make/N8N for workflow automation

API Endpoints Reference

Generate Gamma:

• POST https://public-api.gamma.app/v1.0/generations
• Creates new content from text input
• Returns: {"generationId": "string"}

Create from Template:

• POST https://public-api.gamma.app/v1.0/generations/from-template
• Uses existing Gamma as template
• Returns: {"generationId": "string"}

Receive Generated URLs:

• GET endpoint for retrieving final Gamma URLs after generation
• Check generation status and get shareable links

List Themes:

• GET endpoint returns available theme IDs and names
• Use themeId from response in generation requests

List Folders:

• GET endpoint returns folder IDs for organization
• Use folderIds array in generation requests

Authentication:

• Header: X-API-KEY: <your-api-key>
• OAuth support coming soon (currently API key only)

References

For detailed UI settings guidance, see references/gamma_ui_settings_prescriptive.md For decision matrices on settings, see references/gamma_settings_decision_matrix.md For official API docs, see: https://developers.gamma.app/