What problem does it solve?
Long AI sessions accumulate stale context, repeated reads, and cascading errors that dilute responses and waste tokens; this Skill restores context hygiene and guides splitting large work into session-sized, verifiable chunks so progress stays reviewable and recoverable.
Core Features & Use Cases
- Context hygiene: prefer targeted reads over whole-file dumps, use grepping to locate relevant sections, and read with offsets/limits to minimize context load.
- Session controls: clearly signal context switches and free space with explicit commands such as compacting history for continuity or clearing the session to start fresh.
- Task decomposition: define committable chunks sized to produce reviewable diffs, clear done conditions, and an ordering of dependencies; include a simple decomposition template and commit-oriented handoff steps.
- Use cases: planning a large feature across multiple sessions, recovering from repeated failed attempts, managing long debugging sessions, or when the same file is being read multiple times.
Quick Start
Before starting a large coding task, ask the assistant to decompose the work into committable chunks with clear done conditions and to compact or clear context as needed.
