What problem does it solve? Large, high-risk coding efforts—cross-cutting refactors, unsafe-to-partially-ship migrations, multi-phase architectural changes—often fail because a single agent tries to implement everything at once without structured planning, delegation, or review checkpoints. Deepwork enforces a scheduler-style orchestration model with persistent state and gated reviews. ## Core Features & Use Cases - Persistent Session State: Maintains a task-specific markdown progress file under .slim/deepwork/ capturing goals, research, plans, validation results, and blockers across the session. - Phased Delegation with Review Gates: Splits work into coherent delivery phases, delegates to specialist agents (oracle, librarian, explorer, designer, fixer), and requires an Oracle review gate after each phase with a bounded re-review budget. - Scheduler Discipline: Tracks task/session IDs, waits for background job completion before advancing, and commits at independently valid delivery boundaries. - Use Case: Migrating a codebase to a new module architecture where each phase changes dependency direction—deepwork coordinates research, phased implementation, structure scans, and gated reviews before each commit. ## Quick Start Ask the AI to use the deepwork workflow to plan and execute this large multi-phase refactor with Oracle review gates after each phase.
