executing-plans

Parses staged implementation plans and advances tasks through review and commit cycles.

7|Updated May 13, 2026
One-click install
npx skills add https://github.com/DawnMoon1542/agents-skills --skill executing-plans-dawnmoon1542
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: executing-plans
Source: https://github.com/DawnMoon1542/agents-skills/tree/main/workflow/executing-plans
Command: npx skills add https://github.com/DawnMoon1542/agents-skills --skill executing-plans-dawnmoon1542

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Executing a large multi-stage implementation plan by hand is error-prone: tasks get skipped, progress tracking is lost after context compaction, and incomplete work gets committed. This Skill reads all staged plan files, locates the next unfinished task, and drives each task through implementation, review, and commit in strict order. ## Core Features & Use Cases - Hierarchical Plan Execution: Walks the Stage → Task Group → Task → Step hierarchy serially, re-reading the plan file at each cycle to recover position after context loss. - Pre-flight Plan Review: Validates all stage plans before writing code, checking dependencies, acceptance criteria, and verification commands, and stops on critical gaps instead of guessing. - Strict Completion Gates: Marks a task complete only when all steps run, red-phase failure is verified, target tests pass, and spec plus code-quality reviews succeed. - Use Case: You have a brainstorming index with five staged implementation plans for a refactor. Invoke this Skill to execute every task in order, handing implementation to subagent-driven-development and commits to smart-exec-plan, until final verification passes. ## Quick Start Use the executing-plans skill to read all stage plans from the brainstorming index and start implementing the next unfinished task.

Frequently Asked Questions about executing-plans

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I execute a multi-stage implementation plan with an AI agent?

Organize plans as staged files indexed in a brainstorming document, then have the agent read all plans, locate the first unfinished task, and execute tasks serially. Each task must pass tests and reviews before its progress is marked and committed.

How to resume plan execution after context window compaction?

Re-read the current plan file at the start of every task cycle, since the plan file is the only reliable source of position. Find the first unfinished group, then its first unfinished task, rather than trusting conversation summaries.

When should a coding task be marked complete in TDD workflows?

Mark a task complete only when all planned steps ran, the red-phase failure was verified, target tests pass, and both specification and code-quality reviews pass. There must also be no unexplained changes outside the plan.

Does staged plan execution require each commit to be deployable?

No. Intermediate commits may leave callers unmigrated, builds failing, or services unstartable if the current task finished its planned scope and the remainder belongs to defined later tasks. Full builds and integration tests run after all stages complete.

What should block automated plan execution?

Stop when plans lack key interfaces or acceptance criteria, code conflicts with design in ways you cannot resolve, task dependencies are undefined, red-phase tests cannot be formed, or reviews repeatedly fail. Never guess or modify completed task progress.