freemcplab / Directory / Playground / lean-ctx

● Community yvgude ⚡ Instant

lean-ctx

Name: lean-ctx (Claude Skill)
Author: yvgude

A Rust tool that aggressively prunes context — drops dead imports, trims comments, compresses logs — before sending it to the agent.

lean-ctx is for the cost-conscious: it sits between you and the agent harness, applying token-saving transforms to outgoing context. Works across Claude Code, Codex, Cursor.

▶ Watch demo 📦 Install 💡 Use Cases

Why use it

Key features

Dead-import pruning
Comment trimming (keeps docstrings)
Log compression
Cross-harness
Configurable aggressiveness

Live Demo

What it looks like in practice

ready

Install

Author repository · Claude Skills documentation

Use Cases

Real-world ways to use lean-ctx

Cut session token usage by 40–70%

👤 Claude Code users ⏱ ~15 min intermediate

When to use: Your bills are climbing and you suspect context bloat.

Flow

Install

Wire lean-ctx into your harness.✓ Copied

→ Prefilter active
Measure

Compare token counts before/after.✓ Copied

→ Big drop
Tune

Loosen if accuracy slips.✓ Copied

→ Stable balance

Outcome: Lower bills with same task success.

Combinations

Pair with other MCPs for X10 leverage

lean-ctx-skill + cmux-skill

Watch token graphs while lean-ctx prunes

Combine lean-ctx-skill with cmux-skill: Watch token graphs while lean-ctx prunes✓ Copied

Tools

What this MCP exposes

Tool	Inputs	When to call	Cost
prefilter	(see docs)	Runs over outgoing context	1 call

Cost & Limits

What this costs to run

API quota: N/A
Tokens per call: Saves tokens
Monetary: Free OSS
Tip: Measure quality regressions; aggressive modes can drop important context

Security

Permissions, secrets, blast radius

Credential storage: None

Data egress: None

Troubleshooting

Common errors and fixes

Agent missing context

Lower aggressiveness; whitelist files that matter

Alternatives

lean-ctx vs others

Alternative	When to use it instead	Tradeoff
context-mode	You want sandboxed tool output	Different layer

Resources

📖 Read the official README on GitHub

🐙 Browse open issues

🔍 Browse all 400+ MCP servers and Skills