What problem does it solve?
It prevents inconsistent, untraceable, and poorly structured AI-assisted work by centralizing Datarim’s workflow and storage invariants into a single “load first” ruleset.
Core Features & Use Cases
- Canonical workflow state & storage boundaries: keeps iterative state strictly in datarim/ while long-term archives live in documentation/archive/.
- Thin operational indexes with strict schemas: standardizes tasks.md, backlog.md, and activeContext.md as one-line-per-task pointers rather than full content.
- Deterministic routing and invariants: defines L1–L4 stage routing, command namespace rules (/dr-), archive mapping, and reflection behavior at archive Step 0.5.
- Safety-critical conventions: enforces task ID formats, disallows forbidden directories, blocks overrides for security-critical skills, and mandates path resolution rules.
Quick Start
Load the datarim-system skill first so subsequent fragments correctly apply paths, numbering, backlog/routing rules, and archive/reflection invariants for your current project.