AnthonySmith96

AnthonySmith96 / open-guardian

Public

Stopping smart agents from doing stupid things.

75
11
100% credibility
Found Feb 18, 2026 at 31 stars 2x -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Rust
AI Summary

Open-GuardIAn is a security gateway that filters AI requests to prevent data leaks, prompt injections, and threats while routing to language model providers.

How It Works

1
🔍 Discover the AI Shield

You learn about Open-GuardIAn, a friendly protector that keeps your AI conversations safe from tricks and private info leaks.

2
📥 Bring It Home

Download the ready program and its simple setup files to your computer folder.

3
⚙️ Share Your Preferences

Open the easy settings note and list the AI helpers you use, plus what to hide like emails or secret words.

4
🚀 Wake the Protector

Click to start it up, and it begins guarding your machine quietly and super fast.

5
🔗 Link Your AI Tools

Tell your AI chat or agent app to chat through this protector instead of going straight out.

6
🛡️ Watch It Shine

Send fun or work messages, and see it zap sneaky bad prompts or swap private details with safe placeholders.

🎉 Safe AI Adventures

Now enjoy speedy, worry-free AI help for building bots or chats, knowing everything stays protected.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 31 to 75 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 open-guardian?

Open-Guardian is a Rust-powered reverse proxy that shields AI apps from LLM providers like OpenAI, Groq, or Ollama by scanning prompts for injections, PII leaks, and risky commands like rm -rf. It redacts secrets with context-aware tokens (e.g., ) and routes requests intelligently, all via a drop-in http://localhost:8080/v1 endpoint. Developers get agent-first security that stops stupid actions without crippling legit tool use.

Why is it gaining traction?

It crushes Python gateways on latency—scans hit under 20µs—while adding smart multi-provider routing, zero-trust key injection from .env, and optional local Ollama judge for nuanced calls. TOML config toggles DLP per category, audit modes log threats in JSONL, and CLI handles service install/start/audit. No vendor lock-in; alias models like "fast-model" to Groq for speed.

Who should use this?

Agent builders with AutoGPT or CrewAI preventing infrastructure sabotage. RAG chatbot devs blocking jailbreaks and prompt leaks. Enterprise ops enforcing DLP on outbound LLM traffic without code changes.

Verdict

Promising for open guardian setups, with strong docs, CLI ease, and test coverage, but 19 stars and 1.0% credibility signal early maturity—run your own stress tests first. Ideal if you're stopping smart agents from dumb mistakes.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.