What problem does it solve? Teams struggle to decide what documentation a project actually needs, often writing pages that mirror existing content or guessing at user priorities. This Skill turns a repository's own evidence—CLI surface, issues, changelogs, and real friction-log runs—into one durable, auditable discovery artifact. ## Core Features & Use Cases - Evidence-based discovery: Enumerates every subcommand, entry point, issue title, and changelog entry as candidate tasks, forbidding candidates sourced from existing docs page titles. - Friction logs and user needs: Runs real first-time-persona task attempts with verbatim output, then writes one "As a X, I need to Y, so that Z" need per task with solution-shape rejection. - Typed inventory and IA plan: Types every existing page against a nine-value doc_type enum, maps needs to pages, assigns three tiers (first-steps, everyday, integration), and produces a coverage table, delete list, and nav plan. - Use Case: A maintainer of a CLI tool asks which docs pages to write, merge, or delete before a release; the Skill produces a docs/discovery/use-cases.yaml artifact ranking tasks by a named signal and flagging stub pages for deletion. ## Quick Start Ask the assistant to run a docs discovery on this repository and produce a tiered documentation plan with a page inventory and coverage table.
