executing-planslisted

# Executing Plans Use this skill to execute a written implementation plan directly in the current session. ## Relationship To Context Skills When work is tracked, use `context-task` as the durable execution-record policy and `pi-task` tools for task-record operations: - use `task_read` for the brief, plan, and prior notes - use `task_append_note` for task-specific discoveries and progress - use `task_write` (`document: "plan"`) when a material course correction changes the coherent plan - use `task_write` (`document: "task"`) for stable task metadata or relevant PR references Use `context-memory` only for durable cross-task facts or decisions. ## Why This Skill Exists This is the lighter-weight alternative to `subagent-driven-development`: execute a plan in one session without per-task builder and reviewer dispatches. Use the subagent workflow when stronger context isolation and review gates are needed. ## When to Use This Skill Use it when the plan is approved, clear enough to execute directly, and moderate in size or tightly coupled. Do not use it when the work is highly parallelizable, benefits from strict per-task isolation, or needs repeated review gates. ## Workflow ### 1. Load and review the plan Before editing, read the plan carefully and check for contradictions, missing prerequisites, or unclear steps. For tracked work, read the task brief, plan, and relevant notes with `task_read`. Do not treat a plan as automatically correct just because it exists. ##