# Spec Normalizer > Padroniza documentação existente no formato canônico Spec-Driven. Remove duplicação e melhora rastreabilidade. Use when this capability is needed. - Skill: `tomevault-io/spec-normalizer` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/spec-normalizer` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/spec-normalizer/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/spec-normalizer --- # Spec Normalizer ## Purpose Normalize, restructure, and clean documentation into the canonical spec format: - Reduce duplication - Improve clarity - Enforce consistent templates - Add traceability (PRD → FR → AC → Design) This skill does NOT add new requirements. It reorganizes and clarifies existing information. ## When to Use - User has arbitrary markdown docs, wiki exports, or old PRDs/specs - Plan Lead routes "messy docs" input to this skill ## Inputs - Arbitrary markdown docs, wiki exports, notes - Old PRDs/specs, technical docs ## Outputs - Canonical spec folder structure (spec/) - Updated docs with consistent headings and cross-links ## Rules 1. Preserve meaning. No behavioral changes. 2. If ambiguity exists, annotate instead of rewriting intent. 3. Prefer short, explicit sections with bullets and tables. 4. Add a "Source" note when content was migrated from another doc. 5. If two docs conflict, do NOT resolve—create Open Question(s). ## Process ### Step 0 — Inventory List docs discovered and categorize: Context, Requirements, Design, Decisions, Runbooks. ### Step 1 — Map to Canonical Structure Move/merge content into: spec/00-context/, spec/10-requirements/, spec/20-design/, spec/90-decisions/, spec/99-meta/. ### Step 2 — Remove Duplication If content is duplicated: keep a single source of truth, add links from other places. ### Step 3 — Add Traceability Add links: PRD goals ↔ FRs; FRs ↔ Acceptance criteria; FRs ↔ Design sections (API/data model). ### Step 4 — Quality Pass Check for: Vague terms ("fast", "secure", "user-friendly"); Missing definitions (glossary); Missing unknowns; Overly long paragraphs. ## Canonical Structure Reminder spec/00-context/ | 10-requirements/ | 20-design/ | 30-features/ | 90-decisions/ | 99-meta/ --- > Converted and distributed by [TomeVault](https://tomevault.io/claim/lucasbiason) — claim your Tome and manage your conversions.