Claude Creative and Media Manipulation Skills: A Complete Guide to Visual Generation and Enhancement

Claude offers ten specialized creative and media manipulation skills—including image generation, video processing, design system analysis, and GIF creation—each defined by a SKILL.md file in the ComposioHQ/awesome-claude-skills repository that orchestrates external APIs via the MCP gateway.

The ComposioHQ/awesome-claude-skills repository extends Claude beyond text generation into a comprehensive creative assistant. These skills enable Claude to manipulate visual assets, generate original artwork, download media, and apply professional design themes by following structured instructions stored in dedicated skill folders.

Architecture of Creative & Media Skills

Creative and media manipulation skills in Claude follow a standardized four-layer architecture. Each layer serves a distinct purpose in the execution pipeline, from initial request matching to secure external API communication.

The Four-Layer Stack

• Skill Definition: Each skill resides in its own folder and contains a SKILL.md file with YAML front-matter. This file provides the description, "When to Use" guidance, and step-by-step instructions that Claude follows to execute the task.
• Scripts / Helpers: Optional Python, Node, or Bash utilities perform heavy lifting such as calling image-generation APIs. Claude invokes these through the MCP gateway when the SKILL.md instructions require external computation.
• Resources: Optional reference files including icons, templates, or model weights. These load on-demand to keep skills lightweight.
• MCP Gateway: The secure authentication layer that routes Claude's requests to external services. All skills requiring network access—such as imagen (Google Gemini) and pixelbin-media-generation—traverse this gateway to protect credentials while enabling rich media capabilities.

When users request creative tasks, Claude matches the intent against each skill's metadata in README.md (lines 80-92), then injects the selected skill's instructions into the prompt context.

Complete Catalogue of Creative & Media Skills

The repository currently ships ten distinct skills covering generation, enhancement, and manipulation workflows:

• anydesign/SKILL.md: Analyzes images, URLs, or Figma files and produces comprehensive design system specifications.
• canvas-design/SKILL.md: Generates original visual art (PNG/PDF) from a design-philosophy manifesto.
• imagen/SKILL.md: Integrates with Google Gemini's image generation API to create UI mockups, icons, and illustrations.
• image-enhancer/SKILL.md: Upscales, sharpens, and de-noises screenshots or low-resolution pictures.
• slack-gif-creator/SKISS.md: Produces animated GIFs sized specifically for Slack with optional validators.
• theme-factory/SKILL.md: Applies professional font and color themes to artifacts including slides, documents, and web pages.
• video-downloader/SKILL.md: Downloads videos from YouTube and other platforms in various formats.
• youtube-transcript/SKILL.md: Retrieves and formats YouTube video transcripts for summarization.
• swiftui-design-skill/SKILL.md: Generates SwiftUI UI code and visual mock-ups from textual design briefs.
• pixelbin-media-generation/SKILL.md: Creates and edits images and videos via the Pixelbin API (85+ endpoints).

How to Invoke Creative Skills in Claude

Users trigger these capabilities through natural language prompts. Claude automatically selects the appropriate skill based on intent matching against the SKILL.md metadata.

To enhance a screenshot using the Image Enhancer workflow:

Improve the image quality of screenshot-2024.png

Claude loads image-enhancer/SKILL.md and returns an upscaled, sharpened version.

To generate original artwork via Canvas Design:

Create a poster about "Future-City Urbanism" using the Canvas Design skill.

Claude executes the two-phase Canvas Design workflow defined in canvas-design/SKILL.md: first writing a design-philosophy manifesto, then rendering a .png file.

To upscale a logo to 4K resolution:

Upscale this logo.png to 4K resolution.

This triggers the Image Enhancer "Upscale to 4K" command, returning a high-resolution asset.

To create Slack-compatible GIFs:

Make a 3-second GIF of a cat chasing a laser pointer, size < 5 MB.

Claude runs the Slack GIF Creator skill, applying size validation rules from slack-gif-creator/SKILL.md before returning the file.

To generate UI mockups with Gemini:

Generate a dark-mode login screen with rounded buttons using the imagen skill.

Claude calls the imagen skill, which forwards the prompt through the MCP gateway to Google Gemini's image-generation API and returns a PNG mock-up.

Key Implementation Files

Understanding the underlying SKILL.md files reveals how Claude handles specific media manipulation tasks:

• image-enhancer/SKILL.md: Defines the batch-processing syntax for handling multiple images simultaneously, plus specific parameters for 4K upscaling and de-noising algorithms.
• canvas-design/SKILL.md: Specifies the two-phase workflow where Claude first generates a design-philosophy manifesto, then translates that manifesto into renderable canvas output.
• imagen/SKILL.md: Documents the integration with Google Gemini's API, including prompt engineering guidelines for UI-specific generation.
• slack-gif-creator/SKILL.md: Details strict size constraints (under 5MB) and frame-rate limits imposed by Slack's platform, with validation steps to ensure compliance.
• pixelbin-media-generation/SKILL.md: Enumerates 85+ Pixelbin endpoints available for image transformation and video editing, allowing Claude to construct complex media pipelines.

Summary

• Ten specialized skills cover image generation, enhancement, video processing, and design system extraction.
• Each skill follows the Claude Skill standard using SKILL.md files with declarative metadata and executable instructions.
• The MCP gateway securely authenticates and routes requests to external APIs including Google Gemini and Pixelbin.
• Key skills include image-enhancer for upscaling, canvas-design for generative art, imagen for AI mockups, and slack-gif-creator for animated content.

Frequently Asked Questions

What file format defines Claude's creative and media manipulation skills?

Each skill uses a SKILL.md file containing YAML front-matter that declares the skill's purpose, trigger conditions, and step-by-step workflow instructions. These files reside in repository subdirectories like image-enhancer/SKILL.md and canvas-design/SKILL.md.

How does Claude authenticate to external image generation APIs?

Authentication flows through the MCP gateway, which securely stores API credentials for services like Google Gemini (used by the imagen skill) and Pixelbin. Claude sends requests to the gateway, which then forwards authenticated calls to the appropriate external endpoints without exposing keys in the prompt context.

Can Claude batch process multiple images using these skills?

Yes. The image-enhancer/SKILL.md specifically defines batch-processing syntax that allows Claude to upscale, sharpen, or de-noise multiple screenshots in a single operation, applying the same enhancement parameters across the entire set.

Which skill generates code rather than static images?

The swiftui-design-skill generates SwiftUI code and accompanying visual mock-ups from natural language design briefs. Unlike raster-based skills such as imagen or canvas-design, this skill outputs programmatic UI components that developers can compile directly into iOS applications.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}