What problem does it solve? Raw learnings captured during development accumulate as unstructured inbox entries with no consistent way to classify, score, and promote them into requirements, decisions, designs, or project knowledge. This Skill provides the shared knowledge base that governs the learning pipeline's promote stage. ## Core Features & Use Cases - Schema Definitions: Defines the 13-field inbox entry schema, legacy 5-field mapping, normalization rules, and evaluation-report schema. - Classification and Scoring: Provides problem-class clustering criteria (root cause + recurrence condition + prevention) and an 8-axis weighted evaluation rubric scored out of 40. - Disposition and Prune Policy: Defines 7 disposition categories plus duplicate, adopted-artifact schema, atomic inbox-to-deferred move procedure, and prune rules for the deferred living pool. - Use Case: When running the learning-promote command, the agent loads this Skill to normalize inbox entries, cluster them into problem classes, score each cluster, apply the Decision filtering gate, and generate adopted artifacts routed to backlog-review. ## Quick Start Run the learning-promote command and reference this Skill to classify, score, and promote the entries accumulated in the learning inbox.
