What problem does it solve? Agent conversations lose important decisions, preferences, and deferred options once a session ends, forcing users to repeat context and re-explain choices. This Skill defines clear rules for persisting durable knowledge into the right storage tier without accidentally turning exploratory chat into permanent policy. ## Core Features & Use Cases - Tiered Lesson Storage: Routes knowledge to global reusable skills, workspace memory (.just-demand/knowledge/memory.md), or task-local decisions.md based on reusability. - Deferred Option Tracking: Records robust alternatives that were intentionally postponed, with reason, risk, and revisit triggers, so trade-offs are never lost. - Pre-Archival Extraction: Extracts reusable decisions from tasks before they are archived, preserving knowledge while keeping secrets and one-off details out. - Use Case: When a user picks a quick fix over a robust refactor during a task, the Skill records the deferred option with its risks and revisit conditions so future work can resume the decision. ## Quick Start Record this architectural decision and the deferred robust alternative into workspace memory before archiving the current task.
