AUGMI Skills
A collection of open-source Claude Code skills for content creation, research, media generation, and developer productivity. These skills extend Claude Code with specialized workflows powered by AI APIs (Google Gemini, Fal.ai, Together AI, Replicate, and more).
Quick Start
# Clone into your Claude Code skills directory
git clone https://github.com/kon-rad/augmi-skills.git ~/.claude/skills/augmi-skills
# Or clone individual skills
# Copy any skill folder into ~/.claude/skills/
Once installed, skills are automatically available in Claude Code sessions via slash commands (e.g., /deep-research, /blog-visual-gen:generate).
Skills Catalog
Content Creation
| Skill | Description | Slash Commands | APIs Used |
|---|
| deep-research | Research-to-content pipeline: web research, analysis, multi-format output | /deep-research:research, /deep-research:summarize, /deep-research:analyze, /deep-research:produce, /deep-research:full | Web Search, Twitter/X |
| viral-tweet-crafter | Craft authentic tweets with 8 templates, posting strategy, and quality checks | /viral-tweet:write, /viral-tweet:daily, /viral-tweet:thread, /viral-tweet:react, /viral-tweet:audit | - |
| spec-interview | Interview-driven spec writing for dev tasks with verification criteria | - | - |
Image & Visual Generation
| Skill | Description | Slash Commands | APIs Used |
|---|
| blog-image-gen | Generate images from prompts for blogs, thumbnails, social media | - | Google Imagen 4 |
| blog-visual-gen | Analyze blog content and generate contextual visuals | /blog-visual-gen:generate, /blog-visual-gen:from-file | Google Nano Banana 3 |
| infographic-gen | Research a topic and generate professional infographics | /infographic-gen:generate, /infographic-gen:from-content | Google Nano Banana 3 |
Video & Audio
| Skill | Description | Slash Commands | APIs Used |
|---|
| ai-film-maker | Create personalized 30-second AI short films | - | Fal.ai, Google Veo, Together AI, Replicate |
| youtube-transcript | Extract transcripts from YouTube videos | - | youtube-transcript-api |
| youtube-music-downloader | Download YouTube audio as MP3 | - | yt-dlp |
Document Processing
| Skill | Description | Slash Commands | APIs Used |
|---|
| pdf | Full PDF toolkit: extract, create, merge, split, forms, OCR | - | pypdf, pdfplumber, reportlab |
| large-pdf-reader | Extract text from large PDFs in chunks | - | pypdf, pdfplumber |
Analytics & Reporting
| Skill | Description | Slash Commands | APIs Used |
|---|
| social-analytics | Daily social media analytics collection, time-series tracking, markdown + PDF reports with charts | /social-analytics, /social-analytics:collect, /social-analytics:report, /social-analytics:setup | Postiz API, Puppeteer, Chart.js |
Developer Tools
| Skill | Description | Slash Commands | APIs Used |
|---|
| codebase-tutor | Voice-based codebase learning and quiz sessions | /codebase-tutor:start, /codebase-tutor:quick | Voice Mode MCP |
Skill Details
ai-film-maker
Creates personalized 30-second AI short films. Supports two modes:
- Default: Full creative pipeline from a concept/theme
- Seed: Starts from a user-provided image, analyzed by Gemini Vision
Pipeline: Story Development -> Scene Breakdown -> Image Generation + Face Swap -> Video Generation -> Narration -> Music -> Final Composition
Quality Tiers:
| Tier | Cost | Text-to-Image | Image-to-Video |
|---|
| Cheapest | ~$1-2 | Together AI Qwen | Fal.ai Wan 2.5 |
| Balanced | ~$2-3 | Fal.ai FLUX | Fal.ai Kling 2.5 |
| Highest | ~$8-15 | Together AI Imagen 4 | Google Veo 3 |
Setup:
pip install fal-client replicate requests google-generativeai
brew install ffmpeg
Environment: FAL_KEY, REPLICATE_API_TOKEN, optionally GOOGLE_AI_API_KEY, TOGETHER_API_KEY
blog-image-gen
Generate high-quality images for blog posts, articles, social media, and YouTube thumbnails using Google's Imagen API.
Models: Imagen 4 ($0.02/img), Imagen 4 Ultra ($0.04/img), Imagen 4 Fast ($0.01/img), Gemini 2.5 Flash ($0.02/img)
Features: Single prompt or batch generation from markdown file, multiple aspect ratios (16:9, 1:1, 9:16, 4:3, 3:4), 1K or 2K output
Setup: pip install google-genai pillow
Environment: GEMINI_API_KEY
blog-visual-gen
Analyzes blog post content, extracts key themes, and generates a contextual visual. Unlike simple prompt-to-image, this skill understands your content before generating.
Workflow: Content Analysis -> Theme Extraction -> Custom Prompt Generation -> Image Generation (Nano Banana 3)
Setup: pip install google-genai pillow
Environment: GEMINI_API_KEY
codebase-tutor
Interactive voice-based codebase learning sessions. Maps the codebase, analyzes recent git changes, presents an overview, and quizzes you.
Session Flow:
- Codebase mapping (via cartographer skill)
- Documentation reading (CLAUDE.md, README, codebase map)
- Git history analysis (last 20 commits)
- Voice presentation of overview and recent changes
- Interactive voice quiz (5-7 questions with feedback)
- Performance summary and review suggestions
Requires: Voice Mode MCP server
deep-research
Comprehensive research-to-content pipeline. Takes a topic + optional links and produces analysis and multi-format content.
Pipeline: Research (web + social) -> Executive Summaries -> Pattern Analysis -> Content Production
Outputs: Analysis document, Twitter thread (5-15 tweets), blog post, YouTube script, viral tweets (10), LinkedIn posts (3 variants), image prompts (5-7)
Directory Convention:
INPUT/<YYYYMMDD>/<topic>/sources/ # Raw scraped content
OUTPUT/<YYYYMMDD>/<topic>/ # Final content outputs
infographic-gen
Three-phase infographic generation: research the topic, design the layout with AI, generate the final visual.
Styles: modern, minimal, corporate, playful, technical, dark
Setup: pip install google-genai pillow requests
Environment: GEMINI_API_KEY
large-pdf-reader
Extracts text from large PDFs that exceed normal processing limits by working page-by-page.
Features: Chunked extraction, specific page ranges, organized output with metadata and TOC, OCR fallback for scanned docs
Output:
extracted/
├── metadata.md # Book metadata
├── full-text.md # Complete text
├── chunks/ # Paginated chunks
└── summary.md # Auto-generated TOC
Setup: pip install pypdf pdfplumber
pdf
Comprehensive PDF manipulation toolkit with guides for every common PDF operation.
Capabilities: Text extraction, table extraction (to Excel), PDF creation, merge/split, rotate, watermark, password protection, form filling, OCR for scanned docs
Tools covered: pypdf, pdfplumber, reportlab, qpdf, pdftk, pytesseract
Includes: forms.md (form filling guide), reference.md (advanced features)
spec-interview
Interview-driven spec writer. Asks targeted questions to gather concrete requirements, verification commands, edge cases, and constraints, then produces a structured specification.
Output includes: Requirements list, verification commands, success criteria, automation commands
Supports: Features, TDD, bug fixes, refactors, and multi-phase projects (auto-splits large projects into sequential phases)
viral-tweet-crafter
Configurable tweet crafting skill with 8 tweet templates, posting strategy, and quality checks. Loads user profile/mission context if available, or asks quick questions to establish voice.
Tweet Types: Builder's Log, Manifesto, Teaching Moment, Community Spotlight, Progress Report, Contrarian Take, Personal Story, Thread Teaser
Includes: Weekly posting cadence, engagement principles, quality checklist, hashtag strategy, metrics guidance
Configuration: Define your mission, contrarian bet, values, and voice in a SELF/ directory or answer setup questions on first use.
youtube-music-downloader
Downloads YouTube audio as high-quality MP3 files. Accepts direct URLs, multiple URLs, or search queries.
Setup: brew install yt-dlp ffmpeg
Configuration: Set MUSIC_OUTPUT_DIR env var or uses ~/Music/downloads/ by default
youtube-transcript
Extracts transcripts from YouTube videos with optional timestamps.
Formats: Plain text (default, cleaned into paragraphs) or timestamped ([MM:SS] text)
Supports: youtube.com, youtu.be, embed URLs, raw video IDs, auto-generated and manual captions
Setup: Uses uv run with youtube-transcript-api
social-analytics
Daily social media analytics collection and reporting. Pulls metrics from Postiz across LinkedIn, Instagram, X/Twitter, YouTube, and Facebook. Stores time-series data, generates markdown + branded PDF reports with Chart.js charts, and uploads to Google Drive.
Pipeline: Collect (Postiz API) -> Aggregate (JSON time-series) -> Report (markdown + PDF) -> Upload (Google Drive)
Platforms: Instagram, X/Twitter, LinkedIn (personal + page), YouTube, Facebook
Reports include:
- KPI dashboard: follower counts with daily/weekly/30-day changes
- Follower growth line chart (last 30 days)
- Engagement bar chart (likes/comments/shares by platform)
- Detailed metrics table with engagement rates
- Recent post performance
Data Storage:
data/social-analytics/
├── raw/YYYY-MM-DD/ # Daily API snapshots
├── aggregated/ # Time-series (followers.json, all-metrics.json)
└── reports/ # Markdown + PDF reports
Setup: npm install -g postiz puppeteer, brew install jq, configure config/integrations.json with Postiz integration IDs
Environment: POSTIZ_API_KEY
Environment Variables
| Variable | Required By | Get It From |
|---|
GEMINI_API_KEY | blog-image-gen, blog-visual-gen, infographic-gen, ai-film-maker | Google AI Studio |
FAL_KEY | ai-film-maker | fal.ai |
REPLICATE_API_TOKEN | ai-film-maker | replicate.com |
TOGETHER_API_KEY | ai-film-maker | together.ai |
POSTIZ_API_KEY | social-analytics | Postiz |
MUSIC_OUTPUT_DIR | youtube-music-downloader (optional) | Local path |
Contributing
- Each skill lives in its own directory with a
SKILL.md (or similar) as the entry point
- Python scripts go in a
scripts/ subdirectory
- Keep skills self-contained - no cross-skill dependencies
- Use environment variables for API keys, never hardcode paths
- Include prerequisites, usage examples, and troubleshooting in the skill doc
License
Individual skills may have their own licenses. See each skill's directory for details.
