What problem does it solve? Onboarding a new graduate student requires deciding which research directions get full-pipeline treatment versus limited processing, then propagating those scope decisions consistently across multiple automated cron tasks and persistent memory. Doing this manually risks inconsistent boundaries, wasted compute on out-of-scope papers, and lost convergence decisions. ## Core Features & Use Cases - Scope Tier Classification: Divides research directions into Core (full pipeline: 5 pillars, Synthos, teaching, algorithm components, public dataset analysis) and Peripheral (gap-and-hypothesis only: cornea, lens, vitreous, tear film, tinnitus, concussion biomechanics). - Cron Task Configuration: Configures five scheduled tasks (autonomous-core-researcher, paper-repair, paper-quality-review, paper-layer-b-review, literature-monitor) with explicit allowed and prohibited direction lists, in-scope-only repair, out-of-scope skipping, and tiered report output. - Persistent Convergence Decisions: Writes the scope convergence decision as a durable memory fact so all subsequent cron runs operate from the same state. - Use Case: A student lists "5 pillars public dataset analysis" and "tear film biomechanics" as directions; the skill classifies the first as Core and the second as Peripheral, emits boundary-explicit prompts for all five cron tasks, and records the decision in memory. ## Quick Start Generate an onboarding plan for a new graduate student whose research directions are "5 pillars public dataset analysis" and "tear film biomechanics", including scope tier classification, configurations for the five cron tasks, and a convergence decision written to memory.
