Hey everyone! π
I'm excited to share the biggest update ever for Skill Seekers β the open-source tool that transforms documentation into production-ready AI skills for Claude, Gemini, and OpenAI.
π What's New?
Skill Seekers has evolved from a simple documentation scraper into a complete skill generation factory. You can now create comprehensive AI skills by combining:
- π Web Scraping β Any documentation website (async support for 3x speed)
- π GitHub Analysis β Deep AST parsing for functions, classes, APIs
- π Codebase Analysis β Design patterns, architecture, dependencies
- π PDF Extraction β Tables, OCR for scanned docs, password-protected files
- π Smart Unified Merging β Cross-reference ALL sources with conflict detection
- π― Bootstrap (NEW!) β Generate skill-seekers itself as a Claude Code skill!
β¨ Major New Features
This is the most significant release in Skill Seekers history:
| Feature | Details |
|---|---|
| π― Bootstrap Skill (Self-Hosting!) | Generate skill-seekers itself as a Claude Code skill! Run ./scripts/bootstrap_skill.sh and install to ~/.claude/skills/ |
| π Smart Rate Limit Management | Multi-token GitHub profiles, auto-switching when rate limited, configurable strategies (prompt/wait/switch/fail) |
| π§ Interactive Config Wizard | Beautiful terminal UI for GitHub tokens, API keys, rate limits β run skill-seekers config |
| π¦ Resume Interrupted Jobs | Resume scraping from checkpoints with skill-seekers resume --list |
| Design Pattern Detection | 10 patterns (Singleton, Factory, Observer, Strategy, etc.) with 87% precision |
| Language Support | Python, JavaScript, TypeScript, C++, C, C#, Go, Rust, Java (+Ruby, PHP) |
| Three-Stream Analysis | Code, Docs, and Insights streams for comprehensive skills |
| Architectural Patterns | MVC, MVVM, Clean Architecture auto-detection |
| How-To Guide Generation | Automatically extracts guides from your tests with AI enhancement |
| Config Pattern Extraction | 9 formats (JSON, YAML, TOML, ENV, INI, Python, JS, Dockerfile, Docker Compose) |
| 18 MCP Tools | Use directly in Claude Code, Cursor, Windsurf, VS Code + Cline, IntelliJ |
| 4 LLM Platforms | Deploy to Claude, Gemini, OpenAI, or export as Markdown |
| 1200+ Tests | Production-ready with comprehensive validation |
| MCP Now Optional | Choose your install: pip install skill-seekers (CLI) or skill-seekers[mcp] (full) |
π― NEW: Bootstrap Skill β Self-Hosting!
The coolest feature: You can now generate Skill Seekers itself as a Claude Code skill!
```bash
Generate skill-seekers as a skill
./scripts/bootstrap_skill.sh
Install to Claude Code
cp -r output/skill-seekers ~/.claude/skills/
Now Claude Code knows how to use Skill Seekers! π€―
```
This means Claude can help you create skills... using the skill about creating skills. Meta!
π NEW: SkillSeekersWeb.com
We launched a dedicated website where you can:
- π¦ Browse 24+ Configs β Find ready-to-use configs for popular frameworks
- π Share Your Configs β Contribute and share custom configs with the community
- π Full Documentation β Complete guides for installation, quick start, advanced features
- π One-Click Start β Copy install commands and get started in seconds
Check it out: skillseekersweb.com
π‘ The Magic: Unified Multi-Source Skills
The real power is combining everything:
json { "name": "myframework", "sources": [ {"type": "documentation", "base_url": "https://docs.example.com"}, {"type": "github", "repo": "owner/repo", "code_analysis_depth": "deep"}, {"type": "pdf", "path": "manual.pdf"} ] }
One command. Three sources. One unified skill with: - β οΈ Conflict detection (docs say X, code does Y) - π Documentation gap analysis - π Cross-referenced API information - π Architecture & design pattern insights
π¦ Quick Start
```bash pip install skill-seekers
Scrape docs
skill-seekers scrape --config react
Analyze a codebase
skill-seekers codebase --directory ./my-project
Create unified skill from multiple sources
skill-seekers unified --config my_unified.json
Package & upload
skill-seekers package output/myskill/ ```
π By the Numbers
- β 7.1K+ GitHub stars
- π§ͺ 1,200+ tests passing
- π€ 4 LLM platforms supported
- π¦ 24 preset configs
- π₯ 24 contributors
- π§ 18 MCP tools
π Links
- π Website (NEW!): https://skillseekersweb.com β Browse configs, docs & guides
- GitHub: https://github.com/yusufkaraaslan/Skill_Seekers
- PyPI:
pip install skill-seekers
What skills will you create? I'd love to hear your use cases! Feel free to ask questions or request features. π
[link] [comments]