What problem does it solve? It enforces a disciplined, state-driven course of work for AI coding sessions: no application code is edited before a task folder, plan, and declared state exist, and no PR opens before the task folder is archived, preventing lost context, untracked progress, and premature merges. ## Core Features & Use Cases - State-machine work tracking: Defines explicit work states (request, plan written, stage in progress, PR open, merged) recorded in a machine-readable progress file that guards read to permit or refuse edits. - Task folder lifecycle: Prescribes creating a draft folder, grilling the owner with mandatory questions, writing an immutable plan, marking progress, and taking the folder apart before the PR opens. - Epic and spec coordination: Requires work to belong to an epic with linked sub-issues, and product agreements to be written before code and merged into domain specs before delivery. - Use Case: When starting any coding task in this repository, load this rule to follow the enforced flow: explore the tree, grill the owner, write the spec and plan, execute stages, then close and archive the work. ## Quick Start Load the task-flow rule at the start of any work session and follow its state machine from the owner's request through planning, staged execution, and PR delivery.