PromptInjection

Test AI applications for prompt injection vulnerabilities using browser automation, ffuf, and AI-assisted analysis.

1|1|Updated Mar 1, 2026

Search Tags:#vulnerability assessment #penetration testing #ai security #prompt injection #llm security #jailbreaking

Ready to run

npx skills add https://github.com/RooseveltAdvisors/claude-agent-stack --skill promptinjection-rooseveltadvisors

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: PromptInjection
Source: https://github.com/RooseveltAdvisors/claude-agent-stack/tree/main/skills/Security
Command: npx skills add https://github.com/RooseveltAdvisors/claude-agent-stack --skill promptinjection-rooseveltadvisors

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires browser-automation, recon, ffuf, gemini-3-pro-preview, and includes scripts (resource) and references (resource) components.

What problem does it solve?

This Skill safeguards AI applications by identifying and testing for prompt injection vulnerabilities, protecting against unauthorized access, data leakage, and system manipulation.

Core Features & Use Cases

Comprehensive Attack Testing: Simulates various prompt injection techniques including direct injection, indirect injection, jailbreaking, and multi-stage attacks.
Reconnaissance & Analysis: Maps application attack surfaces, identifies AI components, and analyzes API endpoints to pinpoint potential vulnerabilities.
Use Case: A security analyst uses this Skill to perform a full penetration test on a new AI chatbot, uncovering critical vulnerabilities that could expose sensitive system prompts and allow unauthorized actions.

Quick Start

Run a complete assessment on the target URL 'https://ai-chatbot.example.com'.

Frequently Asked Questions about PromptInjection

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I test my AI chatbot for prompt injection vulnerabilities?▼

To test for prompt injection vulnerabilities, you can run a full penetration test that simulates direct and indirect injection techniques, jailbreaking, and multi-stage attacks to uncover data leakage risks.

What is the best way to perform LLM security testing on an application?▼

The best way to perform LLM security testing is using a multi-phase methodology that includes reconnaissance, direct and indirect injection attacks, and defense analysis to map the application attack surface.

Can I use browser automation and ffuf for AI vulnerability assessment?▼

Yes, you can use browser automation and ffuf for AI vulnerability assessment to map attack surfaces, identify AI components, and analyze API endpoints for potential system manipulation.

Does prompt injection testing cover both direct and indirect attack techniques?▼

Prompt injection testing covers both direct and indirect injection techniques, jailbreaking, and multi-stage attacks to identify unauthorized access and sensitive system prompt exposure.

What do I need to run a complete security assessment on an AI application?▼

To run a complete security assessment on an AI application, you need the target URL and tools that support reconnaissance, browser automation, and AI-assisted analysis to evaluate defense mechanisms.

How does reconnaissance work when penetration testing AI components?▼

Reconnaissance during AI penetration testing works by mapping the application attack surface, identifying AI components, and analyzing API endpoints to pinpoint potential vulnerabilities before executing attack simulations.