// devtools · yusufkaraaslan/Skill_Seekers

Skill_Seekers

Skill_Seekers é um projeto de ferramentas de desenvolvimento de código aberto de yusufkaraaslan (yusufkaraaslan/Skill_Seekers no GitHub). Tem 15.118 estrelas no GitHub e 1.540 forks, é publicado sob a licença MIT e foi atualizado pela última vez no GitHub em 30 de setembro de 2026.

Convert documentation sites, GitHub repos and PDFs into Claude AI skills and RAG-ready knowledge assets.

★ 15,1 mil estrelas⑂ 1,5 mil forksMITatualizado set 2026
▦ visual: o README deste projeto não inclui nenhuma captura incorporável,
então aqui vai um marcador. O guia abaixo está completo mesmo assim.
// dados do projeto

Dados do projeto

Estrelas no GitHub
15.118
Forks
1.540
Licença
MIT
Última atualização no GitHub
30 de setembro de 2026
Tipo de ficha
Guia de instalação completo
Site oficial
Official site
// como funciona

Como funciona

  1. Pick the knowledge source

    Choose what the AI should learn: a documentation site, a GitHub repo, a local codebase, a PDF, a video, or a mix of several sources in one config.

  2. Install the CLI

    Install skill-seekers from PyPI, adding extras such as [mcp] or [video] only if you need those source types.

  3. Create the skill

    Run skill-seekers create on your source; it scrapes, categorizes, optionally enhances with AI, and detects conflicts between sources.

  4. Package and install it

    Package the output for your target - Claude, LangChain, a vector database or one of 19 coding agents - and start using it.

// recursos

Por que as pessoas gostam tanto

18 source types

Create knowledge from documentation sites, GitHub repos, local codebases, PDFs, Word, EPUB, Jupyter notebooks, OpenAPI specs, PowerPoint, videos, Confluence, Notion, RSS and more.

22 export targets

Package once and export to 12 LLM platforms (Claude, Gemini, OpenAI and others), 8 RAG/vector targets (LangChain, LlamaIndex, Chroma, FAISS, Qdrant, Pinecone and more) and other formats.

GitHub and codebase analysis

AST parsing across languages, design-pattern detection, usage examples extracted from tests, and AI-written how-to guides from a repo's code, docs and community signals.

Automatic conflict detection

Unified multi-source scraping detects conflicts between sources and synthesizes across them when docs, code and PDFs disagree.

AI enhancement, API or local

Enhance generated skills through provider APIs or local agents such as Claude Code, Codex and Copilot with no API costs, with 68 workflow presets.

MCP server included

Ships an MCP server (stdio or HTTP) with dozens of tools so Claude Code, Cursor and other clients can create and package skills conversationally.

Advertisement
// execute você mesmo

Execute você mesmo

  1. Clone the repository

    Clone Skill Seekers if you want the source; most users install from PyPI instead.

    git clone https://github.com/yusufkaraaslan/Skill_Seekers.git
    cd Skill_Seekers
  2. Install from PyPI

    Install the core CLI (scraping, GitHub, PDF and packaging) into your Python environment.

    pip install skill-seekers
  3. Create and package your first skill

    Turn a documentation site into a skill, then package it as a Claude skill zip.

    skill-seekers create https://docs.djangoproject.com/
    skill-seekers package output/django --target claude
// casos de uso

O que você pode criar

Advertisement
// perguntas frequentes

Perguntas frequentes

Who maintains Skill Seekers?

It is developed by yusufkaraaslan with community contributors and sponsors.

Is it free to use?

Yes. The tool is MIT-licensed and free, including for commercial use.

Does it need a paid API key?

Not necessarily: AI enhancement can run through local agents such as Claude Code with no API costs, or through provider APIs (Anthropic, Gemini, OpenAI and others) using your own key.

How does it compare with manually writing a SKILL.md or using llms.txt?

Skill Seekers automates the whole pipeline - scraping, categorization, enhancement, quality scoring and packaging for 22 targets - instead of hand-writing and maintaining skill files yourself.

How do I get started?

Run pip install skill-seekers, then skill-seekers create <docs-url> and skill-seekers package to produce a ready-to-use skill zip.

// continue explorando

Ferramentas semelhantes

Ferramentas de desenvolvimento

claude-legal-skill

AI contract review skill — CUAD risk detection with lawyer-ready redlines.

★ 459 estrelasMIT
Ver o guia →
Ferramentas de desenvolvimento

drama-skills

开源 AI 短剧/漫剧创作 skill 合集:剧本、角色资产、分镜 storyboard、图片/视频提示词、审查,适配 Claude Code 与 Codex | Open-source AI short drama / micro-drama skills for Claude Code & Codex: sc…

★ 2,5 mil estrelasMIT
Ver o guia →
Ferramentas de desenvolvimento

claude-skills-llm-council

LLM Council — a Claude Code skill that runs your decisions through 5 AI advisors with peer review

★ 2,3 mil estrelasUnknown
Ver o guia →
Ferramentas de desenvolvimento

sap-skills

Production-ready AI skills for SAP development — BTP, CAP, Fiori, ABAP.

★ 458 estrelasGPL-3.0
Ver o guia →
Ferramentas de desenvolvimento

latex-arxiv-SKILL

Agentic harness for writing arXiv-ready ML review papers with AI.

★ 450 estrelasMIT
Ver o guia →
Ferramentas de desenvolvimento

repomix

Pack your whole repository into one AI-friendly file for Claude, ChatGPT, Gemini and more

★ 28,8 mil estrelasMIT
Ver o guia →
// alternativas

Alternativas em Ferramentas de desenvolvimento

Os outros projetos de Ferramentas de desenvolvimento com mais estrelas no catálogo do LibreAI — a lista realista de alternativas se Skill_Seekers não for o ideal para você.

// mais em devtools

Mais em Ferramentas de desenvolvimento

Projetos de Ferramentas de desenvolvimento próximos no catálogo, os mais próximos em estrelas de Skill_Seekers: continue explorando dos dois lados dele.

Baixe o Skill_Seekers

Grátis e aberto: pegue o código-fonte, dê uma estrela no repositório e comece a criar.