FIELD NOTE / AGENT SKILL
Security Threat Model.
Map assets, trust boundaries, and realistic abuse paths in a repository.
ResearchOpenAISource checked 12 Oct 2026
When it earns a place.
Prioritize mitigations for a specific application or code path.
How to start.
Read the linked SKILL.md as a historical example. OpenAI marks this collection deprecated; follow the OpenAI Plugins repository and current plugin documentation linked from its README for current examples and installation guidance.
Open the original instructionsWhat it needs.
Codex skill with access to the relevant repository and deployment context.
Before you connect it.
Unverified architecture assumptions can distort the threat model; label them explicitly.