Clawland-AI

Clawland-AI / Geneclaw

Public

Self-evolving AI agent framework with 5-layer safety gatekeeper. Agents observe failures, propose fixes, and safely apply them. Built on HKUDS/nanobot.

geneclaw.ai ai-agent autonomous-agent evolution llm nanobot
14
4
100% credibility
Found Feb 23, 2026 at 13 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

Geneclaw extends the nanobot AI agent framework with a self-evolution engine that observes failures, generates safe improvement proposals, and applies them with human oversight and multi-layer safeguards.

How It Works

1
🔍 Discover self-improving AI

You hear about Geneclaw, a smart helper that learns from its mistakes and gets better over time.

2
📦 Set up your assistant

Download and prepare your personal AI companion in a few simple steps.

3
đź’¬ Start chatting

Talk to your assistant like a friend, asking questions or giving tasks, and watch it use helpful abilities.

4
🔧 Turn on learning mode

Switch on the magic that lets your assistant notice its slip-ups and think of ways to improve.

5
đź’ˇ Review smart suggestions

Your assistant shares safe ideas to fix problems it spotted, with checks to keep everything secure.

6
âś… Apply improvements

With your okay, your assistant makes the changes and tests them to ensure they work perfectly.

đź“Š Watch it grow stronger

Check the colorful dashboard to see your assistant getting smarter, with charts of its progress and success stories.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 13 to 14 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 Geneclaw?

Geneclaw is a Python framework that turns lightweight AI agents into self-evolving systems, built on top of nanobot. It records every agent interaction in JSONL logs, analyzes failures heuristically or via LLM, and generates structured code proposals with unified diffs to fix issues. A 5-layer safety gatekeeper—scanning paths, secrets, diff sizes, and risky patterns—ensures proposals are validated before safe application via git branches, pytest runs, and automatic rollbacks.

Why is it gaining traction?

Unlike raw self-evolving agents on GitHub that risk runaway changes, Geneclaw defaults to dry-run mode with explicit human approval, plus autopilot loops for low-risk fixes and a Streamlit dashboard for auditing proposals, timelines, and benchmarks. CLI commands like `nanobot geneclaw evolve --dry-run`, `autopilot`, and `doctor` make iteration dead simple, while event stores with secret redaction keep things secure. It's a practical cookbook for self-evolving agents with reflective abilities, bridging papers and prototypes.

Who should use this?

AI agent builders extending nanobot for production chats on Telegram, WhatsApp, or Slack, who want agents to self-improve from real failures without manual patching. Ideal for indie devs prototyping autonomous retraining pipelines or researchers testing self-evolving LLM agents in controlled sandboxes.

Verdict

Early alpha with 14 stars and 1.0% credibility—docs are solid (runbooks, GEP spec), 123 tests cover core flows, but low adoption means watch for upstream nanobot syncs. Try for self-evolving agent experiments if you're okay bootstrapping; skip for mission-critical agents until more battle-tested.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.