skills.rest
NlightNFotis
NlightNFotisCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

adversarial-design-review

Analyze system designs for adversarial abuse paths and trust boundary failures.

2|Updated Apr 30, 2026
Search Tags:#threat modeling#security review#risk assessment#prompt injection#llm safety#trust boundaries#abuse cases
One-click install
npx skills add https://github.com/NlightNFotis/skills --skill adversarial-design-review-nlightnfotis
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: adversarial-design-review
Source: https://github.com/NlightNFotis/skills/tree/main/adversarial-design-review
Command: npx skills add https://github.com/NlightNFotis/skills --skill adversarial-design-review-nlightnfotis

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill helps you uncover how a design can be abused, bypassed, confused, or quietly gamed before it ships. It is especially useful when normal happy-path reviews miss trust boundaries, malicious actors, prompt injection risks, or incentive-driven misuse.

Core Features & Use Cases

  • Boundary-Centered Threat Modeling: Maps assets, trust boundaries, actors, and invariants so you can see where security and correctness actually change hands.
  • Abuse Case Generation: Uses STRIDE, LLM-specific attack patterns, and incentive failure modes to produce concrete attack scenarios instead of vague warnings.
  • Risk Ranking and Mitigation Design: Prioritizes issues by impact and feasibility, then recommends boundary-level controls, tests, and telemetry.
  • Use Cases: Review authentication flows, plugin systems, tool-calling agents, APIs, secret handling, untrusted file or network inputs, and metrics that could be gamed.

Quick Start

Ask the adversarial-design-review skill to assess your API, CLI, agent, or plugin design by identifying trust boundaries, abuse cases, top-ranked risks, and concrete mitigations.

Frequently Asked Questions about adversarial-design-review

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

FAQPage Schema
How do I find prompt injection risks in my LLM agent design before launch?▼

Threat modeling your LLM agent design identifies prompt injection risks by mapping trust boundaries, enumerating actors, and applying LLM-specific abuse cases to uncover hidden attack paths before release.

What is the best way to perform a security review for a plugin architecture?▼

A security review for plugin architectures requires mapping trust boundaries between untrusted plugins and core systems, then generating STRIDE-based abuse cases to design boundary-level mitigations.

How do you assess trust boundary failures in an API authentication flow?▼

Assessing trust boundary failures in authentication flows involves enumerating actors and invariants, then using impact-feasibility ranking to prioritize security tests and telemetry for boundary-level controls.

Can I use STRIDE threat modeling to find abuse cases in tool-execution pipelines?▼

Yes, applying STRIDE threat modeling to tool-execution pipelines identifies incentive gaming and untrusted input handling risks by generating concrete attack scenarios and recommending boundary-level mitigations.

When do I need an adversarial design review for my CLI or API?▼

Adversarial design reviews are needed when happy-path reviews miss trust boundaries, malicious actors, or incentive-driven misuse in CLIs, APIs, and cross-boundary operations involving secrets, PII, or privileges.

What limitations exist when applying risk assessment to LLM safety designs?▼

Risk assessment for LLM safety requires structured threat modeling with actor enumeration and LLM-specific abuse cases, but cannot guarantee complete coverage of all emergent prompt injection or incentive gaming vectors.

Related Skills

Explore More Software Engineering Skills→
YairamaYairama

security-threat-model

Generate repository-grounded threat models identifying trust boundaries, assets, attack paths, and risks.

Community
Advanced
nguyenpv1980-wqnguyenpv1980-wq
2

threat-modeler

Build security threat models with STRIDE analysis and mitigations.

Community
Advanced
LuisSambranoLuisSambrano
7

code-review

Perform systematic code review and static analysis across multiple programming languages.

Community
Advanced
ntuan2502ntuan2502

security-auditor

Audit code and infrastructure for OWASP Top 10 and compliance vulnerabilities.

Community
Advanced