Skill Basic Scorer

English | 简体中文

An evidence-based Codex skill for manually reviewing, comparing, and scoring Agent Skills.

The included 100-point rubric evaluates:

• discoverability and metadata;
• scope and trigger fit;
• outcomes and success criteria;
• workflow judgment and degrees of freedom;
• prompt economy and progressive disclosure;
• evidence and missing-context behavior;
• output contracts and stop rules;
• validation and benchmark guidance;
• safety and operational guardrails.

Important

This is a custom review heuristic aligned with OpenAI guidance and the Agent Skills specification. It is not an OpenAI certification or official scoring system.

Install as a plugin

Add this GitHub repository as a Codex plugin marketplace:

codex plugin marketplace add Theonefkguy/skill-basic-scorer --ref main

Restart the ChatGPT desktop app, open Plugins, select Theonefkguy Skills, and install Skill Basic Scorer.

Install the standalone skill

In Codex, invoke $skill-installer with:

Install https://github.com/Theonefkguy/skill-basic-scorer/tree/main/plugins/skill-basic-scorer/skills/skill-basic-scorer

Usage

Invoke the skill explicitly:

$skill-basic-scorer Review this SKILL.md

Other examples:

$skill-basic-scorer Compare these two Skill versions.
$skill-basic-scorer Audit every direct child Skill in this directory.

The result separates:

• Agent Skills specification requirements;
• current OpenAI product guidance;
• custom rubric judgment;
• structural validator results;
• live benchmark evidence, when available.

Repository layout

.
├── .agents/plugins/marketplace.json
└── plugins/skill-basic-scorer
    ├── .codex-plugin/plugin.json
    └── skills/skill-basic-scorer
        ├── SKILL.md
        ├── agents/openai.yaml
        └── references/basic-rubric.md

License

MIT