Skill Grading Report: deliberative-refinement - Score 87/100 (B)

[RichardHightower]

✅ Skill Grading Report

Score: 87/100 | Grade: B

Quick Summary

Pillar Scores

Pillar	Score	Max
Spec Compliance	14	15
Progressive Disclosure	26	30
Ease of Use	22	25
Writing Style	7	10
Utility	18	20
Modifiers	+0	±15

Issues Found: 5

• 🔴 High: 0
• 🟡 Medium: 2
• 🟢 Low: 3

📊 Full Grading Report

Skill Evaluation: deliberative-refinement

Score: 87/100 (B)
Repository: aaaronmiller/deliberative-refinement

What's Going On With This Skill?

deliberative-refinement scores 87/100 (B). It works, but there are a few things holding it back from being great.

The Main Issues

Promotional content violates objectivity (Medium priority)

Marketing-style content that doesn't instruct

Fix: Remove or move to README.md; SKILL.md should be purely instructional

Phase 3 numbering inconsistency (Low priority)

Steps jump from 1 to 7 to 8, confusing execution flow

Fix: Renumber steps sequentially: 1, 2, 3

Missing TOC for 130-line file (Low priority)

File exceeds 100 lines but lacks table of contents

Fix: Add brief TOC after the tagline: ## Contents - When to Activate, Core Formula, Execution, etc.

Lacks concrete deliberation example (Medium priority)

No input/output example showing actual deliberation in action

Fix: Add brief example: 'User: Should I use SQL or NoSQL? → Council forms → R1 positions → R2 convergence → Recommendation'

Terminology inconsistency: Probe vs Evidence Grounding (Low priority)

'Probe' and 'evidence grounding' used for same concept

Fix: Standardize on 'Probe' throughout; define once as 'web search for evidence grounding'

Learn More

These recommendations follow Anthropic's Skill Authoring Best Practices. The key insight: Claude picks your skill from 100+ options based on the description alone—so clarity and specificity matter more than you'd expect.

Improvement Checklist

• Promotional content violates objectivity (+2 points)
• Phase 3 numbering inconsistency (+1 points)
• Missing TOC for 130-line file (+1 points)
• Lacks concrete deliberation example (+1 points)
• Terminology inconsistency: Probe vs Evidence Grounding (+1 points)

---

See Skill Authoring Best Practices for guidance.

---

📦 Recommended: Add Universal Installer Instructions

Consider adding these installation instructions to your README.md to help users install this skill across 14+ AI coding agents:

## Use This Skill with Other AI Agents

Ever run out of Claude Code tokens mid-project? Or want to try Gemini CLI or Cursor on a task?

This skill follows the [Agent Skills Standard](https://agentskills.io/), which means you can use it with **14+ AI coding agents**—not just Claude Code. The `skilz` CLI makes this easy.

### Quick Install

```bash
pip install skilz

# Install for Claude Code (your home directory)
skilz install -g https://github.com/aaaronmiller/deliberative-refinement

# Or from the SkillzWave marketplace
skilz install aaaronmiller_deliberative-refinement/deliberative-refinement

That's it. The skill is now available in all your Claude Code projects.

Using with Other Agents

Switch agents with --agent:

# Gemini CLI
skilz install -g https://github.com/aaaronmiller/deliberative-refinement --agent gemini

# OpenCode
skilz install -g https://github.com/aaaronmiller/deliberative-refinement --agent opencode

# OpenAI Codex
skilz install -g https://github.com/aaaronmiller/deliberative-refinement --agent codex

Project-Level Install

Want the skill only in your current project? Add --project:

skilz install -g https://github.com/aaaronmiller/deliberative-refinement --project

Full Platform List

Skilz supports Claude Code, Gemini CLI, OpenCode, OpenAI Codex, Cursor, Windsurf, GitHub Copilot CLI, Qwen Code, Aidr, and more.

See the skill on SkillzWave for agent-specific install commands, or check the skilz-cli docs.

---

SkillzWave is a skill marketplace for AI agents. SpillWave (where I work) builds AI agent tools.

---

**Powered by:** [SkillzWave.ai](https://skillzwave.ai) | [SpillWave](https://spillwave.com)

[rickHasSkilz]

87 is a very high score. Let me know if you have any questions about scoring higher. Good job. We accepted all of your skills. Thanks for using our marketplace. I will write you back when it all goes live. --Rick Hightower