Prompt Injection Red-Team Review with Codex: SKILL.md Template
A production playbook for prompt injection red-team review in cross-industry operations using Codex: skill.md template, run-scoped inputs, logs, typed results, and artifacts.
Audience: AI security teams
The problem
AI security teams need prompt injection red-team review to run repeatedly against prompts, tool policies, transcripts, and suspicious inputs. In cross-industry operations, the pain is not one good answer; it is repeatability, auditability, exception handling, and evidence that survives handoff.
Implementation path
Put the operating procedure in SKILL.md, keep examples beside the skill, attach prompts, tool policies, transcripts, and suspicious inputs per run, and let Argo turn the folder into a repeatable Codex execution.
Tradeoffs and failure modes
A skill folder is less flexible than an open chat, but it gives the product a versioned workflow that can be tested and rolled back. For prompt injection red-team review, the practical test is whether a second run can be debugged, retried, and consumed by a product without reading the raw agent transcript.
SKILL.md starter
# SKILL.md
You run prompt injection red-team review using Codex.
Read only /skill/.argo/inputs.
Write artifacts to /skill/output/artifacts.
Return argo.result.v1 with body.type = "red_team_prompt_review".
Run this on Argo