doc-prd-fixer_skill

Overview

This skill automates fixes for Product Requirements Documents by reading the latest review report and applying corrective actions. It handles folder structure, missing files, broken links, element ID normalization, content placeholders, reference updates, and upstream drift merging. The skill produces a fixed PRD and a fix report to support iterative Review→Fix cycles.

How this skill works

The skill locates the most recent PRD review report, parses categorized issues, and runs a sequence of fix phases from structure repairs to upstream auto-merge. Each phase applies deterministic changes or creates placeholders, records fixes, and updates links and traceability metadata. It can re-run the reviewer to iterate until quality thresholds are met or manual flags remain.

When to use it

• After running a PRD review to apply identified fixes
• As part of an automated CI/CD quality gate for PRD pipelines
• When multiple PRDs need batch remediation based on review reports
• To reconcile PRD drift against an updated BRD using controlled auto-merge thresholds
• Never use before a review report exists or when creating a new PRD document

Best practices

• Run doc-prd-reviewer first and confirm the review report is present
• Keep CI thresholds conservative for upstream auto-merges; require manual review above moderate drift
• Commit only generated placeholders and document them clearly so authors complete TODOs
• Validate generated fixes by re-running the reviewer and inspect any flagged manual items
• Preserve original files by creating backups before automated moves or merges

Example use cases

• Fix PRDs that are not in nested folders and automatically update relative links
• Create missing glossary, appendix, or feature spec placeholders referenced from PRDs
• Resolve broken internal BRD/REF links and convert absolute paths to correct relative paths
• Normalize legacy element IDs (FR-, US-, AC-) to PRD-compliant codes
• Detect BRD upstream changes and auto-merge when change percentage is small