What problem does it solve? When one AI agent works alone on a project, plans drift, progress gets lost between sessions, and reviews are skipped. This discipline keeps a solo agent aligned with a single source of truth for plans, issues, and reviews while preserving resumable context. ## Core Features & Use Cases - Projectman as truth: Plan, issue, and review state live in Projectman; memory only carries resume context. - Bounded microtask flow: Read the active task or sprint, work the smallest useful slice, validate behavior, and log issues for material findings. - Risk-proportional review: Request asynchronous review only when risk warrants it, and write checkpoints at meaningful phase boundaries. - Use Case: An agent resumes a half-finished feature, reads the current sprint, completes one microtask, files an issue for a discovered bug, and writes a checkpoint so the next session can continue cleanly. ## Quick Start Use the solo-pm-loop discipline to continue my current Projectman sprint, complete the next microtask, and write a checkpoint when done.