#169 · Primary category: AI Coding Assistants
lean-ctx
LeanCTX is a local-first context engineering layer for AI coding agents, cutting token usage by 60–90% via compression, caching, and session memory.
Project last updated:08/29/26
GitHub Stars
3.7K
Forks
337
Contributors
56
License
Apache-2.0
Why we included this project
Long coding sessions have a familiar failure mode: the agent burns most of its context window re-reading files and echoing shell output, leaving little room for actual reasoning. LeanCTX is a local Rust binary that sits next to your coding agent and changes that. It decides what the agent reads and compresses file views and command output before the model sees them, which turns cached re-reads from roughly two thousand tokens into about thirteen. Session memory survives across chats, so the agent stops re-reading the same files every turn. An optional local proxy shrinks each request prompt-cache-safe without discarding data, and it works with the usual agents, Cursor and Claude Code included. If long sessions and rising token spend are your bottleneck, this is a direct way to fit more useful context into each dollar spent.
Articles for this project
No articles for this project yet.
To suggest a topic or contribute an article, contact us.
Related projects in this category
superpowers
An agentic skills framework & software development methodology that works.
ECC
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
skills
Skills for Real Engineers. Straight from my .agents directory.
opencode
The open source coding agent.
andrej-karpathy-skills
A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls.