THE MEDIA FOR AI THAT ACTUALLY HELPS.English edition · Understand & take action
OpenAI

Write an autonomous vulnerability ratio, skeptically validated

vulnerability-writeup

In brief

Transforms vulnerability notes, disclosure reports, proof of concept, source code, or Codex Security findings into autonomous vulnerability reports, written in a natural style, and validated with scepticism. It is used for a vulnerability or a disclosure campaign; Codex Security scan is optional.

What it does

  • Writes an autonomous report, based on notes, PoC or observations
  • Validates the statement with skepticism; Codex scan is not mandatory

Who it is for, and an example

Persons who must produce a clear disclosure writing, based on notes, a PoC or findings, for a vulnerability or campaign.

Example request: "From these notes and this PoC, write an autonomous vulnerability report, validated with skepticism. »

The skill is justified in transforming notes, disclosure reports, PoCs, code or Codex Security findings into stand-alone reports for a vulnerability or disclosure campaign. Example written by iadujour.fr from the SKILL.md, not tested.

Install

In Codex: codex plugin add codex-security@openai-curated (Official Place of Deposit).

codex plugin add codex-security@openai-curated

Works with

  • Codex

Source: README: Codex plugin collection; the deposit marketplace is called "Official Codex". Other officers are not appointed.

Level: Advanced

Address for advanced use (security, production, architecture or fine settings) described in SKILL.md.

Good to know

Official deposit OpenAI. These are examples of Codex plugins, often written by the plugin editor (Figma, Stripe, Canva, etc.) and hosted here. Skill plugin "codex-security" in openai/plugins.

SKILL.md excerpt

The original description, which the agent reads to decide whether to load the skill (in English):

Turn vulnerability notes, disclosure reports, PoCs, source code, or Codex Security findings into self-contained, sceptically validated, natural-sounding vulnerability reports. Use for one vulnerability or a disclosure campaign; a Codex Security scan is optional.

Quoted from SKILL.md at commit 0722921 . The folder also contains 2 extra files (agents/, references/).

The SKILL.md has 179 lines and 9 main sections:

  • Purpose
  • Non-negotiable rules
  • Actors, language and release references
  • Trace affected release history
  • Evidence-first intake
  • Campaign workflow
  • Single-finding drafting prompt
  • Source and exploitability standard
  • Report acceptance

Licence: No licence found

Checked on 10 November 2026 : no licence file in the skill folder or at the repository root.

No licence found: short excerpt quoted only (quotation right). For the full text, see the source.

Source. Card written by iadujour.fr from: Frontmatter, sections and body of the deposit. Checked on 10 November 2026. openai/plugins .
SAME PUBLISHER

Other OpenAI skills: Security and identity

Whole category
OpenAISecurity and identity

Manage an API OpenAI key without ever exposing it to clear

openai-platform-api-key

As an input door for identifiers when building, executing, testing, debugging or configuring, in Codex, an application, interface, script, CLI, or generator that relies on OpenAI, or an AI tool whose provider is not specified, including a request formulated only as "using AI".

Tools: CodexLevel: Intermediateupdated 03/11/2026
OpenAISecurity and identity

Create or update a repository threat template

threat-model

Used when Codex is already in the threat modelling phase of a scan, when the person explicitly invokes $threat-model, or when explicitly requests to create, update, or record a repository threat template.

Tools: CodexLevel: Intermediateupdated 08/10/2026
DISCUSSION

Comments.

A field note, a question, a precision on this skill? Comments are reviewed before publication.

Published comments

Loading comments…

    Leave a comment

    Moderation charter. Every comment is reviewed before publication. We refuse hate speech, insulting or discriminatory language, advertising, spam and promotional links, off-topic posts and other people’s personal data.