norika1207-lab

norika1207-lab / afu-brain

Public

OpenClaw-compatible MASL safety gate with public RAG packs for memory-aware AI agents

charenix.comstaticalfred-lobster-brain-demo.html agent-memory agent-safety ai-agents local-first masl
10
3
100% credibility
Found May 08, 2026 at 10 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

Afu Brain provides a local judgment layer for personal AI agents, evaluating requests against safety policies and user memory to approve or block actions before execution.

How It Works

1
🔍 Discover Afu Brain

You stumble upon Afu Brain while looking for ways to make personal AI helpers safer and smarter.

2
👀 Watch the demo

Click the live demo link to see how it carefully thinks before letting AI take important actions like reviewing contracts.

3
🚀 Try the quick test

Run a simple demo on your computer to watch it decide safely on sample tasks, feeling the peace of mind right away.

4
🛠️ Set it up easily

Follow easy steps to get it ready on your machine, using your own sample data without any hassle.

5
📝 Give it a real task

Tell it something like 'review this contract but don't send without my okay,' and see it plan carefully.

6
🧠 See smart judgments

Watch it rank risks, suggest safe steps, and pause for your approval on big decisions, just like a trusted advisor.

✅ Safe AI companion

Now your personal AI thinks twice before acting, learning your preferences for worry-free daily help.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 10 to 10 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 afu-brain?

Afu-brain is a Python safety gate for memory-aware AI agents, compatible with OpenClaw execution layers. It acts as a deterministic decision brain between model proposals and tool actions like emailing or file access, using MASL policies, private owner memory, and public RAG packs to classify risks and enforce approvals. Users get structured JSON decisions—block payments, prepare drafts, or ask for confirmation—plus CLI demos for RAG retrieval, file vault ranking, and evolution traces.

Why is it gaining traction?

It stands out by separating judgment from execution: models suggest, but afu-brain gates irreversible actions with inspectable policies and shared RAG packs for cognition upgrades, without relying on prompts alone. Developers hook into live observatory feeds for evidence-backed improvements, like 93% memory routing accuracy from simulations. Local-first design keeps private data off-cloud while pulling public safety lessons.

Who should use this?

Agent builders integrating OpenClaw tools for personal assistants—think voice butlers handling contracts, receipts, or calendars without blind execution. Indie devs crafting approval-gated workflows for email drafting or file searches. Teams prototyping MASL safety for production agents touching real APIs like Twilio or Google Drive.

Verdict

Early technical preview with solid docs and demos, but only 10 stars and 1.0% credibility signal high risk for anything beyond experiments. Fork and extend if agent safety gates excite you; otherwise, wait for benchmarks and adapters.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.