Skill Seekers

Automated tool that transforms documentation websites, GitHub repositories, and PDF files into production-ready Claude/Cursor skills. Use this skill when the user needs to create, scrape, enhance, package, or upload skills from docs, code, or PDFs.

When to Use This Skill

Trigger when the user:

• Wants to turn a documentation site, GitHub repo, or PDF into a Claude/Cursor skill
• Asks to scrape docs, analyze a repo, or extract content from PDFs for skills
• Needs to package, enhance, or upload an existing skill
• Wants MCP integration, preset configs, or conflict detection between docs and code
• Mentions Skill Seekers, skill-seekers, or "doc to skill" workflows

Tool Overview

Skill Seekers:

1. Scrapes docs, GitHub repos, and PDFs
2. Analyzes code (AST parsing for Python, JS, TS, Java, C++, Go)
3. Detects conflicts between documentation and implementation
4. Organizes content into categorized references
5. Enhances with AI (optional) and packages to ZIP (or platform-specific format)

Result: production-ready skills in ~20–40 minutes instead of manual summarization.

Installation

# CLI only (recommended minimum)
pip install skill-seekers
# or with uv
uv tool install skill-seekers

# With MCP (Claude Code, Cursor, Windsurf)
pip install skill-seekers[mcp]

# Multi-LLM (Gemini, OpenAI)
pip install skill-seekers[all-llms]

# Everything
pip install skill-seekers[all]

Setup wizard: skill-seekers-setup

Core Commands

Command	Purpose
skill-seekers scrape --config <config>	Scrape documentation site (use preset or local JSON config)
skill-seekers scrape --url <url> --name <name>	Quick scrape without config file
skill-seekers github --repo owner/repo	Scrape GitHub repo (code + metadata, optional issues/changelog)
skill-seekers pdf --pdf <path> --name <name>	Extract content from PDF (optional: --ocr, --extract-tables, --parallel)
skill-seekers unified --config <config>	Multi-source: docs + GitHub + PDF with conflict detection
skill-seekers enhance <output_dir>	AI-enhance SKILL.md (API or local agent)
skill-seekers package <output_dir>	Package skill to .zip (default Claude; use `--target gemini
skill-seekers upload <path.zip>	Upload to Claude (requires ANTHROPIC_API_KEY)
skill-seekers install --config <name>	One-shot: fetch config → scrape → enhance → package → optional upload
skill-seekers config --github	Configure GitHub token / profiles (rate limits, multi-account)
skill-seekers install-agent <output_dir> --agent cursor	Install built skill to Cursor (~/.cursor/skills/ when --agent cursor)

Detailed CLI options and platform comparison: see references/cli-reference.md.

Presets and Configs

Presets (by name): react, django, vue, fastapi, godot, ansible-core, etc. Fetch from API or use local JSON in configs/.

skill-seekers scrape --config configs/react.json
skill-seekers install --config react --no-upload

Unified configs (docs + GitHub): e.g. configs/react_unified.json, configs/django_unified.json.

Cursor / Claude Integration

• Install to Cursor: After packaging, run
  skill-seekers install-agent output/<name>/ --agent cursor
  to copy the skill into ~/.cursor/skills/ (or use this project’s /install-skills if the skill lives under skills/).
• MCP: Install with pip install skill-seekers[mcp] and configure the MCP server (stdio or HTTP) for your agent; then use natural language to scrape, package, and manage configs.
• Custom Claude-compatible API: Set ANTHROPIC_BASE_URL and ANTHROPIC_API_KEY for enhancement/upload to non-Anthropic endpoints (e.g. GLM-4.7).

Rate Limits and Profiles

For GitHub scraping, configure tokens to avoid rate limits:

skill-seekers config --github
skill-seekers github --repo owner/repo --profile work   # use named profile
skill-seekers github --repo owner/repo --non-interactive  # CI: fail fast, no prompts

Config: ~/.config/skill-seekers/config.json. Strategies: prompt, wait, switch, fail.

Resume and Large Runs

• Resume interrupted jobs: skill-seekers resume --list then skill-seekers resume <job_id>.
• Async scraping (faster): skill-seekers scrape --config configs/godot.json --async --workers 8.
• Large docs: use skill-seekers estimate <config> first; consider splitting configs and router/hub skills (see official docs).

References

• Repository: github.com/yusufkaraaslan/Skill_Seekers
• Presets and docs: skillseekersweb.com
• CLI and platform details: references/cli-reference.md