Audit SOW/Spec documents against a 100-point scoring framework.
npx skills add https://github.com/thkt/claude-config --skill formatting-audits
Please help me install this Agent Skill. Skill: formatting-audits Source: https://github.com/thkt/claude-config/tree/main/ja/skills/formatting-audits Command: npx skills add https://github.com/thkt/claude-config --skill formatting-audits
SOW/Specドキュメントの品質を100点満点で評価する標準フォーマットを提供します。
/validate でSOW/Specを評価します。結果を参考に改善を進めてください。
High-intent search queries and answers about installing and using this skill.
Use a 100-point scoring framework to audit SOW/Spec documents across four dimensions: accuracy, completeness, relevance, and actionability. Documents scoring 90 points or above pass; results include explicit evidence, risk disclosures, dependencies, and acceptance criteria to guide revision.
A quality spec requires traceable rationale, explicit acceptance criteria, identified dependencies, documented risks, and evidence supporting each requirement. The 100-point audit assesses whether these elements exist and whether reviewers can act on the document without ambiguity.
Run audits during SOW reviews, specification reviews, design reviews, and documentation reviews to surface gaps early, assess quality systematically, and establish a pass/fail threshold before proceeding. The 90-point standard ensures consistency across review cycles.
Yes. The 100-point framework applies to SOW reviews, specification reviews, design reviews, and documentation reviews. The same four evaluation dimensions—accuracy, completeness, relevance, and actionability—work across document types and project contexts.
Documents below the 90-point threshold fail and receive explicit revision recommendations identifying which dimensions—accuracy, completeness, relevance, or actionability—need improvement, along with specific gaps and evidence required to pass revalidation.
bloknayrbAssemble meeting context from notes, emails, action items, and project status.
dwoolworthAutomate QA gate actions on the Planning Board by reading tickets, posting test comments, and updating statuses.
ArroganceGDecompose system design documents into hierarchical WBS tasks with dependencies and acceptance criteria.
HaykTarkhanyanGenerate hierarchical peer-review reports with major and minor comments.
