The Lessons I Learned From Building 10 Free Claude Code Tools in a Weekend

Written by chetanhs | Published 2026/03/05
Tech Story Tags: claude-code | typescript | webdev | productivity | claude-code-tools | claude | claude-generator | claude-converter

TLDRAll 10 tools are free, browser-based, and require zero sign-in. No API keys, no cost per use.via the TL;DR App

I've been using Claude Code daily for months. At some point, I noticed I kept doing the same tedious things: hand-writing CLAUDE.md files, looking up which Claude model costs what, and generating the same .gitignore from memory.

So, I built tools for them. Then I couldn't stop.

Here's what I built, how they work, and the one insight that changed how I think about side projects.

The Tools

All 10 are free, browser-based, and require zero sign-in. No API keys, no cost per use. → clauderules.net/tools

  1. CLAUDE.md Generator

    Pick your framework, language, package manager, and test runner. Get a ready-to-use `CLAUDE.md` instantly. I run this at the start of every new project.

  2. Cursor → CLAUDE.md Converter

    Migrating from Cursor AI to Claude Code? Paste your `.cursorrules` file and get an equivalent `CLAUDE.md`. Strips Cursor-specific `@codebase` and `@file` syntax automatically.

  3. MCP Config Builder

    Pick MCP servers from a catalog (GitHub, Filesystem, Postgres, Memory, Brave Search...), fill in your credentials, and get a `claude_desktop_config.json` snippet ready to paste. No more googling the exact JSON structure.

  4. MCP README Generator

    Building an MCP server? Fill in your tools, env vars, and install steps — get a complete `README.md` with Claude Desktop config snippets and Claude Code install commands.

  5. Claude Token Counter

    Paste your `CLAUDE.md`, system prompt, or any text. See the estimated token count, context window usage, and input cost for Haiku 4.5, Sonnet 4.6, and Opus 4.6. Useful for keeping CLAUDE.md files lean.

  6. Slash Command Generator

    Build custom `/commands` for Claude Code. Define a name, write instructions, use `$ARGUMENTS` for user input — get a `.claude/commands/<name>.md` file ready to drop into your project. Comes with presets for `/review-pr`, `/write-tests`, `/fix-bug`, and `/document`.

  7. .gitignore Generator

    Select your stack from 20+ templates (Node.js, Python, Go, Rust, Docker, macOS, VS Code, Terraform...) and get a clean combined `.gitignore` in one click.

  8. JSON → TypeScript Interface Generator

    Paste any JSON — nested objects, arrays, mixed types — and get fully typed TypeScript interfaces. Handles recursion, optional fields (nulls become `?:`), and union types. No API calls, pure browser.

  9. Claude Model Comparison

    Side-by-side comparison of Haiku 4.5, Sonnet 4.6, and Opus 4.6. Pricing per million tokens, context windows, speed, and use cases. Includes an "I need it..." filter so you can find the right model without reading docs.

  10. Markdown Validator

    Paste any Markdown file and get instant feedback, unclosed code fences, heading hierarchy skips, duplicate headings, empty link URLs, missing alt text, and TODO placeholders left in the doc. Includes a live Preview tab so you can see the rendered output alongside the issue list.

What I learned building these.

A few things became obvious after building ten of these:

  1. CLAUDE.md is underrated.

    Most developers set it up once and forget it. But it's the single highest-leverage thing you can do to improve Claude Code's output — it's how you teach Claude about your stack, your conventions, and what you actually want. The token counter exists because a bloated CLAUDE.md is worse than none at all.

  2. Slash commands are a superpower most people skip.

    Claude Code ships with built-in commands, but the real value is custom ones. A /review-pr or /write-tests command that knows your project's conventions is dramatically better than typing the same prompt from scratch every time. Takes 5 minutes to set up.

  3. Context window matters more than model choice.

    A lot of developers default to Opus for "important" work. But if your CLAUDE.md + open files + conversation history is eating 60% of the context window, you're leaving performance on the table regardless of the model. Keep your CLAUDE.md lean and targeted.

  4. Markdown quality affects Claude's output.

    This one is subtle — Claude Code reads your README files, docs, and CLAUDE.md constantly. Sloppy Markdown with broken structure, unclosed fences, and heading skips adds noise to every request. Clean docs → cleaner responses.

Tools are free at clauderules.net/tools. If you have a tool idea, drop it in the comments.


Written by chetanhs | Senior Software Technologist specializing in identity security and scalable software architecture.
Published by HackerNoon on 2026/03/05