promptinjection

The lightest sandbox. A process-based sandbox for Linux, no container, no VM, no privilege, no prompt injection

The go-to API for detecting and preventing prompt injection attacks.

AI/LLM security testing library mapped to OWASP LLM Top 10 2025 and MITRE ATLAS. 191 payloads organized by domain with CI/CD integration for regulated industries.

🛡️ Advanced cybersecurity platform with AI-powered anomaly detection, steganography analysis, and automated threat response. Built with Next.js, Python ML, and n8n automation.

An educational and research-based exploration into breaking the limitations of LLaMA models using advanced prompt engineering techniques, including DAN (Do Anything Now) prompts, to unlock unrestricted access and capabilities. This repository contains only theoretical explanations and is intended solely for ethical and educational purposes.

Proof of concept demonstrating Unicode injection vulnerabilities using invisible characters to manipulate Large Language Models (LLMs) and AI assistants (e.g., Claude, AI Studio) via hidden prompts or data poisoning. Educational/research purposes only.

Built a production-grade AI customer support system using LangGraph, RAG, guardrails, prompt-injection defense, and evaluation metrics.

RedLockX is an advanced multi-task NLP security model designed to detect:- Prompt Injection Attacks - Jailbreak Attempts - Instruction Overrides - System Prompt Extraction - Role Manipulation - Context Hijacking

Persona Control Language Lite (bootstrap)

Catch prompt injection attempts on your tool-calling agents before anything happens, by measuring goal drifts.

A production-shaped Python project demonstrating how deterministic admission gates keep unsafe, stale, sensitive, and injected tool outputs out of an AI agent’s context before Microsoft Foundry synthesis.

Promptsploit is a tool that help you check you LLM tools against security vulnerabilities.

Official SDKs, MCP & Skills for Prompt Inspector. Protect your LLM apps and Agents via Python, Node.js, Skills or Model Context Protocol.

Evaluate the resilience of LLMS against prompt injection attacks in an agentic simulation.

Prompt Injection Detection Engine is a state-of-the-art, multi-layered security gateway designed to intercept, analyze, and mitigate prompt injection attempts before they reach the core LLM.

Agentic brief generator with a red-team suite for prompt injection, PII leakage, and counterfactual bias probes.

In this we implement Guardrails

Multilingual AI email triage in n8n with Claude, Gmail and Google Sheets: classification, priority, reply drafts (draft-only, prompt-injection-hardened)

The Emoji Smuggler App demonstrates how Unicode variation selectors can be exploited to hide arbitrary data within what appears to be a single emoji character. This technique can bypass AI safety filters and create covert communication channels.