epic-backlog-analyzer

Analyze product backlog structure for epic definition quality, story-to-epic alignment, sizing balance, and thematic coherence across the full backlog. Use when Codex needs to audit backlog health before PI planning, quarterly commitment, or when epic sprawl and story misplacement are suspected.

Epic Backlog Analyzer

Overview

Evaluate backlog-wide epic structure and story placement to surface misalignment before planning. Convert structural debt into a scored health report with concrete restructuring actions.

Workflow

  1. Parse the full backlog CSV and group user stories under their assigned epics.
  2. Evaluate each epic for structural completeness: goal clarity, scope boundaries, outcome metric, and definition of done.
  3. Assess epic sizing against a one-quarter target; flag epics that exceed quarter capacity or are too small to justify epic status.
  4. Analyze story-to-epic alignment by comparing each story theme to its parent epic theme and detecting orphans or misplacements.
  5. Compute per-epic scorecards and an overall backlog health score using weighted dimension scores.
  6. Produce a findings report with a prioritized action table of restructuring recommendations.

Output Contract

Produce one analysis report using backlog-analysis-template.md. Include these sections in order:

  • Backlog health summary with overall score
  • Per-epic scorecards
  • Misalignment findings with recommendations
  • Action table with priority and effort

Quality Gates

Reject or flag output when any condition is true:

  • An epic has no goal statement or definition of done.
  • An epic exceeds 18 stories or 50 story points without a split recommendation.
  • More than 20% of stories in an epic fail theme coherence checks.
  • Orphan stories exist with no parent epic assignment.

Resource Usage

Read epic-health-methodology.md for scoring rubrics and threshold calibration. Use backlog-analysis-template.md for consistent output shape. Run scripts/backlog_health_score.py --input backlog.csv --output backlog-findings.json to compute epic scores, detect misalignments, and produce structured findings.