How does the scoring system work?

The skill evaluates documents on a 0-100 scale, assigning up to 25 points each for Accuracy, Completeness, Relevance, and Actionability.

How do I trigger an audit?

You can trigger the skill by using keywords like 'SOW review', 'Spec review', or 'document review' within your Claude Code environment.

What is considered a passing score?

A score of 90-100 is a PASS, 70-89 is CONDITIONAL (requires fixes), and 0-69 is a FAIL requiring major revisions.

Can I use this for architectural design reviews?

Yes, it is specifically designed for design reviews, SOWs, and technical specifications to ensure they are actionable and comprehensive.

What do the confidence markers mean?

The skill uses ✓ for verified facts, → for inferences, and ? for gaps or unknowns to clarify the evidence behind the evaluation.

Document Formatting Audit & Scoring

Name: Document Formatting Audit & Scoring
Author: thkt

bythkt

•

Learning & Documentation

Evaluates technical specifications and SOW documents using a standardized 100-point scoring system to ensure project readiness.

The formatting-audits skill provides a rigorous framework for evaluating the quality of Statements of Work (SOW) and technical specification documents. By scoring documents across four critical dimensions—Accuracy, Completeness, Relevance, and Actionability—it provides objective feedback on whether a project is ready for implementation. The skill utilizes confidence markers (✓/→/?) to distinguish between factual evidence and inferences, helping teams identify risks, dependencies, and gaps in test coverage before a single line of code is written.

Key Features

01100-point scoring system based on four key quality metrics

02Automated SOW and Spec consistency checking

03Confidence markers (✓/→/?) for evidence-based evaluation

04Pass/Fail/Conditional logic for project phase gates

053 GitHub stars

06Categorized feedback for required vs. recommended modifications

Use Cases

01Reviewing Statements of Work (SOW) to ensure all acceptance criteria are testable

02Validating the alignment between high-level business goals and proposed technical solutions

03Auditing technical specifications for completeness before starting development

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add thkt/claude-config formatting-audits

For use in Claude.ai and ChatGPT

Download Skill