skills.rest
blamejs
blamejsOFFICIAL
Explore all skills by this author →
WebsiteGitHub

threat-modeling-methodology

Enumerate AI threats and map them to ATLAS, ATT&CK, CWE, and CVE catalogs.

Updated May 11, 2026
Search Tags:#threat modeling#compliance#stride#ai security#linddun#cwe mapping#mitre atlas
One-click install
npx skills add https://github.com/blamejs/exceptd-skills --skill threat-modeling-methodology
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: threat-modeling-methodology
Source: https://github.com/blamejs/exceptd-skills/tree/main/skills/threat-modeling-methodology
Command: npx skills add https://github.com/blamejs/exceptd-skills --skill threat-modeling-methodology

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Most threat models in use as of mid-2026 are outdated, failing to account for modern AI-specific threats including AI agent actors, MCP supply chain risks, prompt injection as an access control bypass, and vector embedding data stores. Global security and AI frameworks also do not mandate current threat modeling methodologies, leading to compliance gaps, paper-only models, and unaddressed attack surfaces for AI and cloud-native systems.

Core Features & Use Cases

  • Methodology Selection & Composition: Guides selection and combination of STRIDE, PASTA, LINDDUN, Diamond Model, MITRE Unified Kill Chain v3.0, and AI/agent-specific composite methodologies to match a system's unique threat surface, avoiding thin coverage from single-methodology use.
  • Threat Mapping & Scoring: Enumerates threats per selected methodology, maps them to ATLAS, ATT&CK, CWE, and known CVEs, and scores relevant threats with RWEP (not CVSS alone) for accurate risk prioritization.
  • Compliance & Operational Validation: Includes a compliance theater check to ensure models are co-located with the systems they cover, include AI actors, and meet cross-jurisdictional framework requirements, plus handoffs to downstream skills for currency assessment and mitigation mapping.
  • Use Case: A team building an AI agent system with MCP plugins and RAG retrievers uses this skill to produce a threat model that covers agent trust boundaries, maps prompt injection risks to ATLAS TTPs, and meets EU AI Act, NIS2, and ISO 42001 requirements.

Quick Start

Use the threat-modeling-methodology skill to build a current, compliant threat model for your AI agent system that includes MCP server trust boundaries and maps all identified threats to relevant ATLAS and ATT&CK techniques.

Frequently Asked Questions about threat-modeling-methodology

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

FAQPage Schema
How do I build a threat model for AI agent systems that covers MCP supply chain risks?▼

To build a threat model for AI agent systems, select a composite methodology like STRIDE or PASTA, enumerate threats across agent trust boundaries, map them to ATLAS and ATT&CK techniques, and apply RWEP risk scoring to prioritize MCP supply chain vulnerabilities.

What threat modeling methodology works best for prompt injection and AI-specific threats?▼

A composite threat modeling methodology combining STRIDE, LINDDUN, and the MITRE Unified Kill Chain works best for AI threats, mapping prompt injection access control bypasses to ATLAS TTPs and CWE catalogs to ensure accurate risk representation.

How do I map enumerated threats to ATLAS, ATT&CK, and CWE catalogs for compliance?▼

Mapping enumerated threats involves selecting methodologies like PASTA or the Diamond Model, identifying specific AI attack vectors, and correlating them to ATLAS, ATT&CK, CWE, and known CVEs to satisfy cross-jurisdictional compliance requirements for EU AI Act, NIS2, and ISO 42001.

Does threat modeling for AI systems require RWEP risk scoring instead of CVSS?▼

Threat modeling for AI systems benefits from RWEP risk scoring over CVSS alone, as RWEP provides more accurate risk prioritization for modern AI-specific threats like agent actors and vector embedding data stores when mapping to ATLAS and ATT&CK.

Why do traditional threat models fail to cover modern AI threats and compliance gaps?▼

Traditional threat models fail because they do not account for AI agent actors, MCP supply chain risks, or prompt injection bypasses, and they lack mapping to ATLAS or compliance checks for frameworks like the EU AI Act, resulting in unaddressed attack surfaces.

Can I use STRIDE and LINDDUN together for cloud-native infrastructure threat modeling?▼

You can combine STRIDE and LINDDUN methodologies for cloud-native infrastructure threat modeling to match a system's unique threat surface, avoiding thin coverage from single-methodology use and ensuring comprehensive mapping to ATT&CK and CWE catalogs.

Related Skills

Explore More Legal & Compliance Skills→
sanity-iosanity-io
4

shape-your-agent

Craft system prompts for Sanity Agent Context MCP agents.

Official
Intermediate
emilneuraz-aiemilneuraz-ai

threat-mitigation-mapping

Map identified threats to appropriate security controls for risk-based mitigation planning.

Community
Intermediate
robotijnrobotijn
3

ai-governance-checker

Scan software repositories for missing AI regulatory compliance artifacts.

Community
Advanced
h4vzzh4vzz
33

threat-modeling

Identify and document security threats using STRIDE, DREAD, PASTA, and attack trees.

Community
Advanced