What problem does it solve? Business knowledge locked in functional specs, ADRs, meeting notes, and interviews is hard to query, easy to misremember, and loses provenance when summarised. This Skill curates those sources into a structured knowledge/ corpus where every fact carries an honest basis (documented, stated, inferred, assumed) and a stable ID, so downstream analysis and spec-driven work can cite exactly where each fact came from. ## Core Features & Use Cases - Structured decomposition: Routes content into typed artifacts — rules, workflows, journeys, glossary terms, constraints, data schemas, integrations, configurations, and UI specs — filed under platform, domain, and feature tiers with fixed ID conventions. - Elicitation-first method: Asks questions in chat one at a time instead of guessing, records open questions as OQ files, and never self-verifies content (everything stays status: draft until a human confirms it). - Workspace scaffolding and hygiene checks: An additive setup script scaffolds the knowledge/ tree without overwriting, and four read-only check scripts validate frontmatter, ID placement, example-content leakage, and tier structure. - Use Case: Point the Skill at a 60-page functional specification; it chunks the document along its table of contents, confirms a business-language topic grouping with you, extracts rules and workflows into domains/<domain>/features/<feature>/, updates the coverage and data-ownership registries, and flags every gap it could not resolve. ## Quick Start Ask your agent to curate the source document at docs/functional-spec.docx into the knowledge base, confirming the domain and feature when prompted.
