Skip to content

Linux endpoint agent for privacy-first monitoring of AI/LLM usage. Part of the Red Specter Purple Team AI Defense Suite.

License

Notifications You must be signed in to change notification settings

RichardBarron27/redspecter-ai-usage-watchdog

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

30 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Red Specter Logo

🛡️ Red Specter – AI Usage Watchdog

Part of the Red Specter Purple Team AI Defense Suite

License Platform Language Version Status

A privacy-first Linux agent for monitoring AI usage on endpoints with zero content capture.


🔍 Description

AI Usage Watchdog is a lightweight Linux-first endpoint agent that detects and logs AI / LLM usage on the host in a privacy-preserving manner.

It complements other Red Specter AI Security Suite tools:

Tool Purpose
AI Endpoint Guard Enforces safe AI usage rules
AI Breach Monitor Detects abnormal AI behaviour
AI Usage Watchdog Visibility & governance

Together, these form the Red Specter Purple Team AI Defense Suite.


✨ Features

Feature Status
AI usage detection (runtime signatures)
JSONL logging for SIEM ingestion
Lightweight agent
Zero prompt/content capture
Local event dashboard (watchdog_dashboard.py) 🚧 v0.2
Policy engine + fleet aggregation ⏳ v0.3

🚀 Quick Start

Install dependency

sudo apt install python3-psutil
Run a single scan (debug)
cd agent
./redspecter_ai_usage_watchdog.py --once --debug

Run continuously
./redspecter_ai_usage_watchdog.py --interval 15

Logs stored at
~/.redspecter_ai_watchdog/logs/events.jsonl


Each event includes:

Timestamp

PID + process name

Username & hostname

Matched signature (risk/category)

Privacy posture:

❌ No prompt text

❌ No file/document contents

✔ Only metadata captured

🔍 Live TUI Dashboard (v0.2)
cd tools
./watchdog_dashboard.py


Features:

Auto-refreshing (every 3s)

Color-coded severity

Keyboard navigation

q → quit

r → force refresh

A lightweight SOC-style view into real-time AI usage on the endpoint.

🗺 Roadmap
Version	Goal	Status
v0.1	Agent MVP	✔
v0.2	Dashboard viewer + SIEM export	🚧
v0.3	Policies + central aggregation	⏳
v1.0	Public hardened release	⏳

See ROADMAP.md for full intent & backlog.

🎯 Purple Team Mission

Offense-driven defense.
We use adversarial insight to design better defensive controls.

Always authorized. Always ethical.
Red Specter defends by thinking like an attacker.

❤️ Support Red Specter
---

## ⭐ Support & Contribute

If you find this project useful, please consider:

- ⭐ Starring the repository — it helps others discover the project
- 🐛 Reporting issues and ideas via GitHub Issues

👉 PayPal: https://paypal.me/richardbarron1747

Every contribution helps build more tools like this — thank you!

Notice for Users: If you cloned this and found it useful, please consider starring the repo! Stars help with visibility and let me know which projects to maintain.