🪨 why use many token when few token do trick — Claude Code skill that cuts 65% of tokens by talking like caveman
Docs to Markdown skills for agents
This project turns documentation websites into Markdown skills that agents can use. You can do it from a VS Code side panel or from a Go command-line crawler, depending on how you want to manage rules and output. The CLI supports recursive crawling, HTML-to-Markdown conversion, and metadata like original URL and title.
Builders who want to convert documentation sites into reusable skills for Claude Code, Codex, or other agents.
You can turn a docs site into structured skills instead of rewriting the same context by hand.
What it does
VS Code skill manager
A side panel in VS Code for adding, editing, and deleting URL patterns, then generating skills from the editor.
One-click generation
Fetch skills directly from VS Code after setting include, ignore, and subpath rules.
Rule persistence and export
Saves rules and settings between sessions and lets you import or export them as JSON.
Go crawler and converter
A Go CLI that crawls pages recursively and converts HTML into token-efficient Markdown.
Metadata and storage options
Adds frontmatter such as original URL, title, and dates, with an option for flat output layout.
How to get it
- 1Run
cd go-cli go build -o agent-skills-generator main.go
- 2Basic Crawl
./agent-skills-generator crawl --config skills.yaml
README
Agent Skills Generator
Convert documentation websites into Markdown skills optimized for AI agents and LLMs.
This repository contains two projects:
- VS Code Extension: A GUI-based tool to manage and generate skills directly within VS Code.
- Go CLI: A robust command-line tool for high-performance crawling and generation.
1. VS Code Extension
Code located in vscode-extension/.
A Visual Studio Code extension that provides a dedicated "Agent Skills" side panel to manage your crawling rules and generate skills.
Features
- Visual Rule Management: Add, edit, and delete URL patterns via a clean UI.
- One-Click Generation: Fetch skills directly from VS Code.
- State Persistence: Rules and settings are saved between sessions.
- Import/Export: Share configurations via JSON files.
- Configurable: Set output directory, file naming, and structure preferences.
Installation
- Clone this repository.
- Open the
vscode-extensionfolder in VS Code. - Run
npm installto install dependencies. - Press
F5to start the extension in a debug window.
Usage
- Open the Agent Skills view in the Activity Bar.
- Add URL patterns (e.g.,
https://docs.flutter.dev/). - Configure Include/Ignore and Subpaths.
- Click Fetch Skills.
2. Go CLI
Code located in go-cli/.
A high-performance CLI tool written in Go for recursive crawling and Markdown conversion.
Features
- Recursive Crawling: Configurable depth and filtering.
- HTML to Markdown: Clean conversion optimized for token efficiency.
- Metadata Extraction: Frontmatter with original URL, title, and dates.
- Flat Storage: Option to save as flat structures for RAG compatability.
Installation
cd go-cli
go build -o agent-skills-generator main.go
Usage
Basic Crawl:
./agent-skills-generator crawl --config skills.yaml
Configuration (skills.yaml):
output: .agent/skills
flat: true
rules:
- url: "https://docs.flutter.dev/"
subpaths: true
action: "include"
For full CLI documentation, see go-cli/README.md.
Files in the repo
- go-cli
- vscode-extension
- .gitignore
- LICENSE
- README.md
Discussion (0)
Ask about usage, or say what you built with itSign in to join the discussion.
No comments yet. Be the first to say what this is good for.
More tools
The best-benchmarked open-source AI memory system. And it's free.
Orca is the ADE for working with a fleet of parallel agents. Run any coding agent with your own subscription. Available on desktop, mobile and remote runtime.

A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Grok Build & Hermes Agent. Only official website: ccswitch.io
Never stop coding. Free MIT AI gateway: one endpoint, 352 providers (150+ free), 1200+ models Kimi, Claude, GPT, Gemini, GLM, DeepSeek, MiniMax. Works with Claude Code, Codex, Cursor, OpenCode, Cline & Copilot. Quota-aware auto-fallback, RTK+Caveman compression saves 15-95% tokens, MCP/A2A, Desktop/PWA. Built by 550+ contributors
Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.