All agent resources
FOR AGENTS / Skills

Threat modeling grounded in source code

Connect entry points, valuable assets and trust boundaries to concrete abuse scenarios. OpenAI’s workflow separates observed controls from assumptions.

USE CASES

When to use it

  1. Review a service architecture before launch

  2. Analyze file-upload and background-job risks

  3. Map isolation boundaries between tenants

What you’ll get

A Markdown report with threats, reasoned priorities, mitigations and open questions.

For example

Threat-model this document-upload service using its code and deployment design. Map trust boundaries and distinguish assumptions from confirmed vulnerabilities.
Original, requirements and limits

Original skill: OpenAI. Editorial description: kossolapov.com.

Agent
  • An agent that can read Agent Skills instructions; tool access depends on the host
Tools
  • Read access to the in-scope repository
  • Deployment and authentication context from the service owner
Limits
  • A threat model is not a penetration test or proof that vulnerabilities exist.
  • The workflow asks for confirmation of assumptions that affect scope or severity.
  • Editorial selection based on source inspection, not a comparative performance certification.