yvgudeyvgudeCommunityยท1 Agent Skills Included

lean-ctx

Cut AI coding agent token costs with smart context compression

Compresses file reads, shell output, and search results before they reach your AI coding agent, cutting token usage by 50-80%. Eliminates wasted context from repeated file reads, verbose git and build logs, and lost session memory between chats. Works locally with Cursor, Claude Code, Copilot, Codex, and 30+ agents, with a live dashboard to track token savings and costs.
npx skills add yvgude/lean-ctx --all -g -y
Available:

Instructs the AI agent to route all file reads, searches, and shell commands through lean-ctx compression tools, register on the local agent bus, and record decisions and progress for session continuity.

All Skills in This Repository (1)

Pure Emerald Level Indicators

Frequently Asked Questions

FAQPage Schema
How to install lean-ctx?โ–ผ

Run `npx skills add yvgude/lean-ctx --all -g -y` in your terminal to install the skill globally, then run `lean-ctx setup` to wire it into your coding agent.

How does lean-ctx reduce AI token costs?โ–ผ

It compresses file reads, shell output, and search results before they reach the model, cutting token usage by 50-80%. Cached re-reads of unchanged files cost only about 13 tokens.

Which AI coding agents work with lean-ctx?โ–ผ

It works with Cursor, Claude Code, Copilot, Windsurf, Codex, Gemini, and over 30 other agents with zero configuration required.

Does lean-ctx keep my code and data private?โ–ผ

Yes. It runs entirely locally on your machine, and your context and session memory stay portable and never leave your control.

Can lean-ctx remember context between chats?โ–ผ

Yes. Its session memory persists tasks, decisions, and findings across chats, so your agent never loses track of previous work.

Related Repositories in Software Engineering

View All in Software Engineeringโ†’