exec-plan

A self-contained execution plan is a living document that guides stateless agents or developers through complex work without prior context. You need one for multi-phase initiatives requiring repeatable, auditable progress and traceability.

Create a new plan under the plans/ directory named after your task, then populate milestones with discrete, idempotent steps, explicit inputs, expected outputs, and exact acceptance criteria to guide autonomous agents or developers.

Milestone-driven plan management handles unknowns by enforcing spike handling for technical discoveries and maintaining a decision log with rationale and rollback guidance, ensuring resilience and recovery during complex software project execution.

Yes, the execution plan is portable and usable by humans or agents across contexts without reengineering. It provides exact commands and explicit acceptance criteria, ensuring stateless agents and developers can execute complex work consistently.

The best way to track progress is using a living document that evolves with discoveries, maintaining a decision log for rationale and rollback guidance. This ensures traceability and repeatable, auditable progress throughout the workflow.

Idempotent steps and validation are needed in plan management to ensure repeatable execution without side effects. By defining discrete steps with exact commands and expected outputs, the plan enforces auditable progress and safe recovery.