Skillkit
A curated collection of AI coding skills for creative and productivity workflows. Each skill is a standalone markdown file that works with any AI coding assistant.
Skills
| Skill | What it does | |-------|-------------| | Image Fetcher | Context-aware image search from Unsplash, Pexels, Pixabay. Zero-config with WebSearch fallback. | | Logo Generator | Adaptive brainstorming, SVG logo creation, full brand kit export (favicons, social images, PWA manifest). | | Presentation Chef | Apple Keynote-style HTML presentations with cinematic animations, glassmorphism, and 5 themes. Single .html file. | | Video/Audio Processor | Transcribe recordings with Whisper, extract visual frames with ffmpeg, synthesize meeting notes. Cross-platform with hardware-aware model selection. | | Local Image Gen | Generate images locally with Stable Diffusion. Auto-detects hardware (Apple Silicon, NVIDIA, AMD, CPU) and picks the best model. Cross-platform. | | Icons8 | Fetch icons, illustrations, and photos from Icons8's 1.5M+ asset library. API key or free CDN fallback. | | Creative Code Lab | Generate interactive generative art: particle systems, flow fields, fractals, shaders, fluid simulations. | | Premium Hero Builder | Agency-quality hero sections with cinematic video backgrounds, glassmorphism, and editorial typography. | | Raycast Automator | Build Raycast extensions and script commands from natural language descriptions. Full lifecycle scaffolding. |
Installation
<details open> <summary><strong>Claude Code</strong></summary>
Install the entire collection as a plugin:
/plugin marketplace add sacredvoid/skillkit
/plugin install skillkit@sacredvoid-skillkitOr install individual skills by copying the SKILL.md file to ~/.claude/skills/<skill-name>/SKILL.md.
Then invoke any skill:
/image-fetcher hero image for a landing page
/logo-generator
/presentation-chef
/video-audio-processor
/local-image-gen abstract neural network on dark gradient</details>
<details> <summary><strong>Cursor</strong></summary>
Copy the skill's SKILL.md into your project as a Cursor Rule:
.cursor/rules/<skill-name>.mdcAdd frontmatter:
---
description: "<skill description>"
alwaysApply: false
---</details>
<details> <summary><strong>Windsurf</strong></summary>
Copy SKILL.md to .windsurf/rules/<skill-name>.md and set activation mode to Manual or Model Decision.
</details>
<details> <summary><strong>GitHub Copilot</strong></summary>
Copy SKILL.md to .github/instructions/<skill-name>.instructions.md with frontmatter:
---
applyTo: ""
---</details>
<details> <summary><strong>Cline / Continue / Aider / Zed / Other</strong></summary>
Each skill is a standalone markdown file. Copy the SKILL.md into your tool's instruction/rule format:
- Cline:
.clinerules/<skill-name>.md - Continue:
.continue/rules/<skill-name>.md - Aider:
aider --read skills/<skill-name>/SKILL.md - Zed:
.zed/prompts/<skill-name>.md - Any tool: Paste or reference the SKILL.md contents as custom instructions
</details>
Skill Details
Image Fetcher
Scans your project directory for context (README, package.json, CSS) and generates relevant search queries. Fetches from Unsplash, Pexels, and Pixabay via API keys or zero-config WebSearch fallback. Presents curated picks with descriptions, sizes, and source info before downloading.
Config (optional): skills/image-fetcher/config.json for API keys.
Logo Generator
Interactive workflow: context scan, adaptive brainstorming (skips questions it can infer), 2-3 concept proposals, SVG generation (light/dark/mono variants), and full brand kit export.
Output: 3 SVG variants, 7 favicon PNGs + ICO, Apple touch icon, OG + Twitter social images, PWA manifest, and HTML preview.
Requires: Node.js for the export script (cd skills/logo-generator/scripts && npm install). Falls back to macOS native tools if sharp is unavailable.
Presentation Chef
Guided workflow: content source, theme selection (with AI recommendations), slide structure review, then generation. 5 themes (Dark Cinematic, Light Minimal, Warm Editorial, Neon Cyberpunk, Earth Organic), 13 slide types, speaker notes, PDF export.
Output: A single self-contained .html file with zero dependencies.
Video/Audio Processor
Extracts visual frames with ffmpeg (1 per minute), transcribes audio with Whisper (local, no API key), and synthesizes findings into structured notes. Auto-detects your hardware and picks the optimal transcription engine: faster-whisper for NVIDIA GPUs, standard openai-whisper for Apple Silicon and CPU.
Requires: ffmpeg and Python. Install commands provided per platform (macOS, Linux, Windows).
Local Image Gen
Generate images locally using Stable Diffusion. Detects your hardware (Apple Silicon MPS, NVIDIA CUDA, AMD ROCm, CPU) and recommends the best model and resolution. Supports SDXL-Turbo, SD-Turbo, and Tiny-SD.
Requires: Python and PyTorch. Install commands provided per platform.
Icons8
Search 1.5M+ icons across 116 styles, vector illustrations, and 500K+ stock photos from Icons8. Works with an API key for full access or free CDN fallback for icons (no key needed).
Config (optional): skills/icons8/config.json for API key and defaults.
Creative Code Lab
Generate interactive generative art and creative coding sketches. Supports particle systems, flow fields, fractals, L-systems, noise-based visuals, text effects, fluid simulations, and interactive 3D scenes. Output as self-contained HTML, React component, Next.js page, p5.js sketch, or raw Canvas/WebGL.
Premium Hero Builder
Build agency-quality hero sections and landing pages with cinematic video backgrounds, liquid glass morphism, editorial typography, and pixel-perfect dark UI. Based on the Viktor Oddy methodology for premium web design.
Raycast Automator
Build Raycast extensions and script commands from natural language task descriptions. Handles complexity analysis, code generation, TypeScript/React scaffolding, and dev launch. Automatically picks script command vs. full extension based on the task requirements.
