Score Backlog

Overview

Read backlog items from a file, issue tracker export, or user input, then score and rank them using a structured prioritization framework (RICE by default, ICE or custom weights on request). Every score includes transparent rationale per dimension so the team can audit, debate, and adjust priorities with shared context instead of gut feelings. Items with low confidence are flagged for further research rather than ranked with false precision.

Workflow

1. Read product context — Load .chalk/docs/product/0_product_profile.md, existing research syntheses, JTBD canvases, and OSTs. These inform scoring — an item connected to a validated customer need gets higher confidence than one based on a hunch. If no product context exists, note this and proceed, but flag that scores will be lower confidence without research backing.

2. Load backlog items — Parse $ARGUMENTS to identify the source of backlog items:

   • A file path (e.g., .chalk/docs/product/backlog.md, a CSV, or a list in markdown)
   • Inline items provided by the user in their message
   • A glob pattern or folder to scan

   For each item, extract: title, description (if available), and any existing metadata (priority, labels, estimates). If items lack descriptions, work with titles only but note that scores will be less reliable.