tmolavi

Lean Agent Skills — AI skill for Claude Code

AI community

⚡ Cut AI coding agent token usage without losing quality — curated Agent Skills for Claude Code, Codex, Antigravity, Cursor, Copilot & more.

How to install Lean Agent Skills

This entry records only its repository, not the path inside it, so there is no exact command to give. Open tmolavi/lean-agent-skills and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Lean Agent Skills does

⚡ Cut AI coding agent token usage without losing quality — curated Agent Skills for Claude Code, Codex, Antigravity, Cursor, Copilot & more. One-command install.

Alternatives in AI

  • Codeburn — Free, local tool to track AI coding token usage and cost across 37 tools and agents (Claude Code, Cursor, Code 9.7k ★
  • Jcodemunch MCP — Cut AI token costs 95%+ on code exploration 2.6k ★
  • Token Monitor — Local-first desktop widget for tracking token usage, costs, and limits across 32+ AI coding tools—including Cl 1.8k ★

README

⚡ Lean Agent Skills

Your coding agent is re-reading your entire codebase every turn — and you're paying for it.

**One command installs curated, battle-tested Agent Skills that slash token usage of AI coding agents — without lowering output quality.**

[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE) [![Standard: Agent Skills](https://img.shields.io/badge/standard-Agent%20Skills-blue)](https://agentskills.io) [![Agents supported](https://img.shields.io/badge/agents-10%2B-success)](#supported-ai-coding-agents) [![Curated skills](https://img.shields.io/badge/skills-45%2B-orange)](#curated-skill-catalog)

[Quick start](#-quick-start) · [Why it saves tokens](#-why-this-saves-tokens) · [Supported agents](#-supported-ai-coding-agents) · [Catalog](#-curated-skill-catalog) · [FAQ](#-faq) · [فارسی](README.fa.md)


😤 The problem

Every AI coding agent — **Claude Code, OpenAI Codex, Google Antigravity, Cursor, GitHub Copilot, Gemini CLI** — burns tokens on the same three things:

  1. Re-discovery. Each session, the agent re-explores your repo to learn conventions it already learned yesterday.
  2. Bulky always-on context. Giant rule files (AGENTS.md, CLAUDE.md, .cursorrules) get resent on every single turn.
  3. Trial and error. Vague tasks → exploration → wrong turns → fix-up rounds. Each round resends the whole conversation.

You feel it as: hitting context limits, slow turns, and a token bill that keeps growing.

✅ The fix

Agent Skills are the open standard ([agentskills.io](https://agentskills.io)) adopted by **40+ clients** — including Claude Code, Codex, Copilot, Cursor, Antigravity and Gemini CLI. A skill is a folder with a `SKILL.md`; agents load them **progressively**:

Mechanism What happens Token effect
Progressive disclosure Only each skill's name + description (~50–100 tokens) sits in context; full instructions load only when a task matche