skills.rest
devCharuzu
devCharuzuCOMMUNITY
Explore all skills by this author →
GitHub

ai-security

Scan prompts for injection signatures and map findings to MITRE ATLAS techniques.

Updated Apr 16, 2026
Search Tags:#jailbreak#atlas#ai-security#risk-scoring#prompt-injection#security-assessment
One-click install
npx skills add https://github.com/devCharuzu/philfida-taskmanage --skill ai-security-devcharuzu
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ai-security
Source: https://github.com/devCharuzu/philfida-taskmanage/tree/main/.windsurf/skills/ai-security
Command: npx skills add https://github.com/devCharuzu/philfida-taskmanage --skill ai-security-devcharuzu

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

This skill provides methodology and tooling for AI/ML security assessment—scanning prompts for prompt injection signatures, scoring model inversion risk, data poisoning exposure, and agent tool abuse. It maps findings to MITRE ATLAS techniques and recommends guardrail controls. It clarifies its focus on AI/ML system security rather than general application security.

Core Features & Use Cases

  • AI Threat Scanner Tool: Scans prompts for injection signatures, scores risk, and maps findings to MITRE ATLAS techniques.
  • Prompt Injection Detection & Jailbreak Assessment: Identifies direct and indirect prompt injection, jailbreak phrases, and system prompt extraction attempts.
  • Model Inversion & Data Poisoning Risk: Assesses inversion risk by access level and data poisoning risk across fine-tuning scopes.
  • MITRE ATLAS Coverage & Guardrails: Provides ATLAS mappings and pragmatic guardrail design patterns to mitigate findings.
  • Cross-References & Workflow Guidance: Helps integrate findings into threat detection, incident response, and cloud-security contexts.

Quick Start

Run the threat scanner on your seed prompts to detect prompt injection and jailbreak risks.

Frequently Asked Questions about ai-security

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

FAQPage Schema
How do I detect prompt injection and jailbreak vulnerabilities in LLM applications?▼

Prompt injection detection scans prompts against predefined static signatures to identify direct and indirect injection, jailbreak phrases, and system prompt extraction attempts, then scores the risk to map findings to MITRE ATLAS techniques.

Can I map AI security threats to MITRE ATLAS techniques during pre-production testing?▼

Yes, MITRE ATLAS mapping is applied during static assessment and pre-production testing to categorize identified threats, scoring risks like agent tool abuse and mapping results directly to guardrail recommendations for mitigation.

How do I mitigate agent tool abuse risks in AI systems?▼

Agent tool abuse risks are identified during security assessment and mitigated by mapping the findings to pragmatic guardrail design patterns and control recommendations that restrict unauthorized tool execution.

Does this AI security assessment cover general application security vulnerabilities?▼

No, this assessment focuses exclusively on AI and ML system security, specifically targeting prompt injection, data poisoning, model inversion, and tool abuse threats rather than general application security vulnerabilities.

Related Skills

Explore More Software Engineering Skills→
NlightNFotisNlightNFotis
2

reverse-engineering

Probe opaque systems to build confidence-rated behavioral models.

Community
Advanced
drupadsachaniadrupadsachania

mitre-atlas

Map MITRE ATLAS threats to AI/ML systems across data pipelines and endpoints.

Community
Advanced
msaad00msaad00
29

agent-bom-analyze

Analyzes AI infrastructure blast radius and attack-path risk using Python 3.11+ and public CVE databases.

Community
Advanced
lebsrallebsral
11

ai-testing-safety

Identify AI safety vulnerabilities through automated adversarial testing with DSPy.

Community
Advanced