Caua-ferraz

Caua-ferraz / AgentGuard

Public

AgentGuard is a firewall for AI agents, preventing that any unwanted surprises go without supervision by your agent

12
0
100% credibility
Found Mar 29, 2026 at 12 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Go
AI Summary

AgentGuard provides policy enforcement, real-time monitoring, and logging to safely control actions taken by autonomous AI agents.

How It Works

1
🔍 Discover safe AI helpers

You hear about powerful AI assistants that can do tasks like browsing or running commands, but worry they might go too far, then find AgentGuard to add safety.

2
📥 Get the safety guard

Download the ready safety tool and start it on your computer with simple steps.

3
✏️ Set your safety rules

Create easy rules in a guide file, like blocking dangerous file changes or needing okay for big actions.

4
🚀 Launch the protector

Turn on the guard service and open the dashboard to see everything in real time.

5
🔗 Link your AI assistant

Connect your AI helper so it asks permission for every action before doing it.

6
📊 Watch and approve

Use the dashboard to review actions live, approve tricky ones, and check logs anytime.

✅ Safe automation achieved

Your AI now works reliably on tasks without risks, giving you full control and peace of mind.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 12 to 12 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 AgentGuard?

AgentGuard is a Go-built firewall acting as an agent guardian for AI agents, providing runtime verification of AI agents to block unwanted surprises without constant supervision. It proxies agent actions through YAML policies covering shell commands, filesystem access, network calls, browser navigation, and cost limits, with options to deny, allow, or require human approval. Users run it via CLI or Docker, monitor via a live web dashboard, and integrate with Python/TypeScript SDKs plus adapters for LangChain, CrewAI, and browser-use.

Why is it gaining traction?

Simple YAML rules with per-agent overrides and rate limiting make policy tweaks fast, while approval queues trigger Slack or webhook alerts for risky moves like sudo commands. Full audit logs and CLI queries let you replay sessions without "it worked on my machine" headaches. Go's performance ensures low-latency checks, and framework adapters mean zero refactoring for popular agent setups.

Who should use this?

AI engineers at startups deploying LangChain or CrewAI agents that run shell tools or hit APIs, needing quick guardrails against prod accidents. Teams automating research bots or deploy scripts, where supervision prevents file deletions or quota blowouts. Devs prototyping agentguard ai flows before scaling to production.

Verdict

Early with 12 stars and 1.0% credibility score, but mature docs, CLI like "agentguard audit," and a packed roadmap make it a strong prototype pick for agentguard runtime verification of ai agents. Hold off for prod until more adoption if stability is critical.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.