Claude Flow AI Defence

Detect prompt injection, prevent jailbreaks, and identify PII in content.

1|Updated Feb 8, 2026
One-click install
npx skills add https://github.com/ricable/cli-skills-builder --skill claude-flow-ai-defence
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: Claude Flow AI Defence
Source: https://github.com/ricable/cli-skills-builder/tree/main/.claude/skills/claude-flow-aidefence
Command: npx skills add https://github.com/ricable/cli-skills-builder --skill claude-flow-ai-defence

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill protects against AI manipulation by detecting prompt injection, preventing jailbreaks, identifying Personally Identifiable Information (PII), and analyzing threats with self-learning capabilities.

Core Features & Use Cases

  • Prompt Injection Detection: Identifies and neutralizes attempts to manipulate AI behavior.
  • Jailbreak Prevention: Stops users from bypassing AI safety protocols.
  • PII Detection: Scans content for sensitive personal information.
  • Threat Analysis: Learns from new threat patterns to improve defenses.
  • Use Case: When processing user-generated content, use this Skill to ensure it's safe and hasn't been tampered with before it reaches your AI model.

Quick Start

Use the Claude Flow AI Defence skill to defend against AI manipulation by running the security defend command.

Frequently Asked Questions about Claude Flow AI Defence

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

FAQPage Schema
How do I detect prompt injection in user-generated content before processing it?

To detect prompt injection, you can scan user-generated content to identify and neutralize manipulation attempts before the text reaches your AI model. This ensures the input is safe and has not been tampered with.

What is the best way to prevent jailbreaks in AI applications?

The best way to prevent jailbreaks is to analyze user inputs against learned threat patterns to stop users from bypassing AI safety protocols. This approach uses vector search for advanced detection of manipulation tactics.

Can I scan for PII and analyze content safety simultaneously?

Yes, you can scan for PII and analyze content safety simultaneously by running a security defense command. This identifies sensitive personal information while detecting AI manipulation threats within the same workflow.

Does AI threat analysis support self-learning capabilities for new attack patterns?

AI threat analysis does support self-learning capabilities by continuously analyzing new threat patterns to improve defenses. This allows the system to adaptively enhance its detection of prompt injection and jailbreak attempts.

How does vector search work for advanced AI manipulation detection?

Vector search enables advanced AI manipulation detection by comparing input content against learned threat patterns in a vector space. This method identifies subtle prompt injection and jailbreak attempts that traditional rule-based filters might miss.