amazinglvxw

amazinglvxw / enso-os

Public

Enso — A lightweight (1267 LOC) self-evolving harness that makes AI agents learn from mistakes. 10 Shell Hooks. Just bash + python3.

agent-memory agent-os ai-agent claude-code harness-engineering
20
0
100% credibility
Found Apr 08, 2026 at 20 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Shell
AI Summary

Enso is a simple add-on harness for AI coding assistants that automatically captures errors, distills lessons, injects them into future sessions, enforces self-protection, and prunes outdated knowledge.

How It Works

1
🔍 Hear about Enso

You find this tool that helps your AI helper learn from its mistakes so it doesn't repeat them.

2
📥 Get it set up

You download it and run a quick setup that takes just 30 seconds on your computer.

3
🚀 Start chatting with your AI

Your AI assistant is now wrapped in a smart safety net that watches everything automatically.

4
❌ AI makes a mistake

During a task, your AI hits an error, and Enso quietly notes it down without you lifting a finger.

5
💡 Next time, it remembers

In your next conversation, the AI avoids the same problem because Enso taught it the lesson.

🧠 Your AI keeps improving

Over time, your helper gets smarter, forgets useless old info, and stays safe—all on its own.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 20 to 20 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is enso-os?

Enso OS is a 1267 LOC bash and Python3 harness that wraps AI agents like Claude Code, intercepting tool calls to capture errors, distill 1-3 atomic lessons per session, and inject them automatically next time. It solves the problem of agents repeating mistakes by enforcing a code-driven learning loop with active forgetting—stale lessons decay after 37 days, LRU caps at 50 entries. Install in 30 seconds via `bash install.sh`; no npm, pip, or Docker needed—just hooks into your agent's lifecycle.

Why is it gaining traction?

Unlike Claude Code's silent truncation or Hermes Agent's growing skills, Enso forces learning via immutable hooks the agent can't bypass, adds quality linting for duplicates/orphans, and self-protects core files. Developers dig the grep-searchable simplicity and local-only storage in `~/.enso/`, plus DIKW distillation (data to wisdom) without RL complexity. At 20 stars, it's niche but hooks tinkerers seeking Claude Code upgrades without bloat.

Who should use this?

Claude Code users grinding terminal sessions on git ops, file IO, or browser automation, tired of manual error fixes. Backend devs building enso ai github tools or laravel enso github workflows who want agents that evolve from failures. Not for multi-model setups needing UI—stick to terminal agents with hook support.

Verdict

Try Enso OS if you're on Claude Code and value transparency over features; its 1.0% credibility reflects 20 stars and early maturity, but solid docs and dogfooding make it low-risk for experiments. Pair with host compression for production—great bootstrap for self-evolving agents.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.