skills.rest
rocklambros
rocklambrosCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

threat-modeling-agentic-systems

Map MAESTRO, MITRE ATLAS, or OWASP catalog items to agentic trust boundaries.

2|Updated May 23, 2026
Search Tags:#threat modeling#agentic ai#maestro#owasp asi#memory poisoning#mitre atlas#runaway loop
One-click install
npx skills add https://github.com/rocklambros/rcs --skill threat-modeling-agentic-systems
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: threat-modeling-agentic-systems
Source: https://github.com/rocklambros/rcs/tree/main/skills/security/threat-modeling-agentic-systems
Command: npx skills add https://github.com/rocklambros/rcs --skill threat-modeling-agentic-systems

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill helps you systematically threat-model autonomous and multi-agent AI systems before deployment so you can catch agent-specific risks early instead of after incidents.

Core Features & Use Cases

  • Catalog-based analysis: Works against a user-supplied framework such as MAESTRO, MITRE ATLAS, or OWASP Agentic Security Initiative.
  • Agent boundary mapping: Evaluates planner-to-executor flows, tool-result feedback loops, memory persistence, loop control, identity scope, and agent-to-agent communication.
  • Practical security review: Useful for code-review agents, research agents, workflow orchestrators, and any tool-using system with iterations, scratchpad memory, or multi-agent collaboration.

Quick Start

Ask the skill to threat-model your agentic system using a specific catalog and include the agent’s tools, memory model, loop limit, and identity scope.

Frequently Asked Questions about threat-modeling-agentic-systems

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

FAQPage Schema
How do I threat model an agentic AI system before deployment?▼

Threat model agentic AI systems by mapping catalog items from frameworks like MAESTRO or MITRE ATLAS to agentic trust boundaries, rating likelihood and impact, and producing a threat register with mitigations and coverage gaps.

What security risks do multi-agent workflows and tool-using assistants face?▼

Multi-agent workflows face risks like memory poisoning, runaway loops, and agent-to-agent communication exploits across planner-executor flows, tool-result feedback loops, and identity scope boundaries.

Can I use OWASP Agentic Security Initiative to review planner-executor agents?▼

Yes, you can apply the OWASP Agentic Security Initiative to review planner-executor agents by supplying the catalog to evaluate agent boundaries, loop control, memory persistence, and identity scope.

How do I analyze the blast radius of a runaway loop in autonomous agents?▼

Analyze runaway-loop blast radius by evaluating the agent's loop limits, tool-result feedback loops, and scratchpad memory within the threat modeling process to contain potential autonomous execution failures.

What do I need to provide to start threat modeling an autonomous AI system?▼

Provide a specific threat catalog like MITRE ATLAS, MAESTRO, or OWASP ASI, along with details about the agent's tools, memory model, loop limits, and identity scope to map agentic trust boundaries.

Related Skills

Explore More Software Engineering Skills→
antgroupantgroup
34

secureclaw

Audit AI agent environments for misconfigurations and apply hardening rules.

Official
Advanced
GuicedEEGuicedEE
5

senior-secops

Triage security incidents and generate containment playbooks for on-prem and cloud environments.

Official
Intermediate
mcauduro0mcauduro0
1

Agent Designer - Multi-Agent System Architecture Toolkit

Design multi-agent system architectures and generate tool schemas.

Community
Advanced
ShaheerKhawajaShaheerKhawaja
8

pos-security

Audit codebases for OWASP Top 10, dependency CVEs, and secret patterns.

Community
Advanced