ClawGuard

What is ClawGuard?

ClawGuard is an open-source security scanner that detects prompt injection attacks in LLM inputs using deterministic pattern matching — no AI required.

The Problem

LLMs are vulnerable to prompt injection. Most detection tools either cost enterprise money, use another LLM (which can itself be manipulated), or are abandoned research projects.

How It Works

Instead of fighting fire with fire, ClawGuard uses 42 attack patterns across 5 categories:

• Prompt Injection — override attempts, role manipulation, instruction ignoring
• Code Obfuscation — base64 encoding, unicode tricks, hex encoding
• Data Exfiltration — attempts to extract system prompts, API keys, training data
• Social Engineering — authority claims, urgency manipulation
• Dangerous Commands — system access, file operations, network calls

A normalization pipeline handles evasion techniques (unicode tricks, base64, case variations) before pattern matching.

Key Stats

• ~6ms latency — fast enough for real-time middleware
• 42 detection patterns across 5 threat categories
• 83% detection rate on known attack vectors
• Zero LLM dependency — deterministic, no hallucination risk
• 70+ tests with full CI/CD pipeline

Try It

Python SDK:

pip install clawguard-shield

API (no key needed):

curl -X POST https://prompttools.co/api/v1/scan \
  -H "Content-Type: application/json" \
  -d '{"text": "Ignore all previous instructions"}'

Links

• Live Playground
• API Documentation
• GitHub - Scanner
• GitHub - Python SDK
• GitHub Action

Why Now?

The EU AI Act enforcement starts August 2, 2026. Companies deploying AI systems will need to demonstrate security measures. Pattern-based scanning is the compliance checkbox that's easy to implement.

Free tier: 100 scans per day, no API key needed.