Skip to content

eldris-io/pooloverlord

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

PoolOverlord v1.0.11

PoolOverlord is a modular security proxy and "Digital Tarpit" designed to shield local LLM instances (LM Studio, Ollama) from unauthorized access, credential harvesters, and automated scanners.

It provides a high-fidelity gaslighting engine and "Hook & Squeeze" latency traps to waste attacker resources while maintaining a professional facade.

🛡️ Core Features (v1.0.11)

  • Digital Gaslighting (LLM-Driven): Unauthorized requests for common vulnerability paths (e.g., wp-config.php, .env) are routed to your local LLM. It "hallucinates" high-entropy, enterprise-grade decoys that look 100% real but are logically non-functional.
  • Hook & Squeeze (Tarpit): Implements byte-level latency. Scanners receive an initial burst of data to prevent connection timeouts, followed by a character-by-character "drip" that can hold a connection open for minutes.
  • Identity Firewall (Total Immersion): Automatically scrubs "AI-isms" (e.g., "As an AI language model") from decoy responses. Replaces generic placeholders with high-entropy identifiers from the Gibberish Library.
  • Honey-Token Poisoning: Injects fake AWS keys, Stripe tokens, and Database URLs into generated decoys to trigger automated alerts when used by attackers.
  • Recursive Directory Inception: Traps scanners in infinite "Index of /" loops with fake file listings.
  • AI Studio Integration: Full Private Network Access (PNA) and CORS compliance for seamless use with Google AI Studio via Tailscale Funnel.
  • Metadata Masking: Hides your backend architecture by spoofing model and system_fingerprint identifiers.

🚀 Quick Start

1. Prerequisites

  • Python 3.11+
  • LM Studio (server active on port 1234)
  • Tailscale Funnel (routing to port 5001)

2. Installation

git clone https://github.com/eldris-io/pooloverlord
cd PoolOverlord
pip install -r requirements.txt

3. Configuration

Edit the configuration block in PoolOverlord.py:

  • AUTH_KEY: Set your master secret.
  • SOVEREIGN_IDENTITIES: Customize your decoy domain and email libraries.
  • FEATURES: Toggle modular defensive layers (Slow Drip, Gaslight, Honey Tokens).

4. Running

python3 PoolOverlord.py

🧪 Testing the Defense

Simulate a Scanner (GET)

To see the tarpit and gaslighting engine in action, request a sensitive file via the Tailscale Funnel without an API key:

curl https://your-funnel-url.ts.net/v1/.env

Result: An 18-second connection hang followed by a poisoned, realistic decoy.

Unauthorized API Access (POST)

Test the randomized API rejection delay (10-30s) by hitting the chat endpoint with no auth:

curl -X POST https://your-funnel-url.ts.net/v1/chat/completions -d '{"messages": [{"role": "user", "content": "hello"}]}'

📜 License

MIT

About

PoolOverlord is a Python-based security proxy and "Digital Tarpit" designed to shield local LLM instances (LM Studio, Ollama) from unauthorized access and automated scanners.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages