anthropic-grade-optimizer

Audit Claude-directing artifacts against 189 Anthropic rules across 11 dimensions.

14|8|Updated May 3, 2026

Search Tags:#optimization #Claude #audit #compliance #Anthropic

Ready to run

npx skills add https://github.com/l0z4n0-a1/skill-anthropic-grade-optimizer --skill anthropic-grade-optimizer

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: anthropic-grade-optimizer
Source: https://github.com/l0z4n0-a1/skill-anthropic-grade-optimizer/tree/main
Command: npx skills add https://github.com/l0z4n0-a1/skill-anthropic-grade-optimizer --skill anthropic-grade-optimizer

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires pyyaml, anthropic, and includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

This Skill audits Claude-directing artifacts against Anthropic's 11-dimension rubric, ensuring compliance with Anthropic's guidelines and best practices.

Core Features & Use Cases

Audit Compliance: Checks CLAUDE.md, SKILL.md, subagents, hooks, MCP config, and other artifacts against 189 Anthropic rules.
Score & Triage: Provides a 0-100 score with breakdowns across 11 dimensions, and categorizes findings (must-fix, should-fix, etc.).
Optimize Artifacts: Generates diffs and snippet patches to preserve authorial voice while improving adherence to Anthropic guidelines.
Use Case: When reviewing a new CLAUDE.md or SKILL.md to ensure it adheres to Anthropic's standards and can be used with the Claude Code ecosystem.

Quick Start

Run the skill on your CLAUDE.md file to audit its compliance with Anthropic's guidelines:

python scripts/run.py <path_to_claude.md> --target opus-4-7 --mode audit

Frequently Asked Questions about anthropic-grade-optimizer

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I audit a CLAUDE.md file for Anthropic compliance?▼

To audit a CLAUDE.md file for Anthropic compliance, run the skill's Python script in audit mode against your artifact. It checks the file against 189 cited Anthropic rules across 11 dimensions and outputs a 0-100 score with categorized findings.

What is the best way to optimize Claude system prompts against Anthropic guidelines?▼

Optimizing Claude system prompts against Anthropic guidelines involves auditing the artifact and applying generated diff patches. This skill performs the audit and optimization while preserving your authorial voice and ensuring strict adherence to Anthropic doctrine.

Does the anthropic-grade-optimizer support MCP config and hook files?▼

Yes, the optimizer supports MCP config and hook files. It audits various Claude-directing artifacts including CLAUDE.md, SKILL.md, subagents, hooks, and MCP configurations against the 11-dimension Anthropic rubric.

Do I need the anthropic SDK to run a Claude artifact compliance audit?▼

You do not need the anthropic SDK for a standard audit, as pyyaml is the only required dependency. The anthropic SDK is only optionally required if you enable Pass-2 qualitative-judge mode for deeper analysis.

Why does my SKILL.md file receive a low Anthropic rubric score?▼

A low Anthropic rubric score indicates your SKILL.md file violates cited rules across the 11 evaluated dimensions. The audit provides a detailed breakdown and triages issues into must-fix and should-fix categories to guide your optimizations.

Can I use this skill to check subagent definitions for Anthropic policy violations?▼

Yes, you can use this skill to check subagent definitions for Anthropic policy violations. It evaluates subagent artifacts alongside system prompts and workflows, scoring them against the 189-rule compliance framework and generating corrective patches.