executing-plans

Executes written implementation plans with task tracking, verification, and review checkpoints.

Updated Aug 3, 2026
One-click install
npx skills add https://github.com/m-de-graaff/skills --skill executing-plans-m-de-graaff
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: executing-plans
Source: https://github.com/m-de-graaff/skills/tree/main/skills/executing-plans
Command: npx skills add https://github.com/m-de-graaff/skills --skill executing-plans-m-de-graaff

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Written plans often get executed carelessly: steps are skipped, verifications are assumed rather than run, and silent deviations make the final result unpredictable. This Skill enforces disciplined execution of an existing plan so the outcome matches what was designed, with honest reporting when it doesn't. ## Core Features & Use Cases - Critical plan review before coding: Reads the entire plan first, flags contradictions, missing prerequisites, and stale assumptions in one message before any work begins. - Task-level verification discipline: Creates a task per plan item and refuses to mark anything complete until its specified verification has actually run and passed. - Blocked vs. plan-defect handling: Distinguishes a blocked step (stop and report) from a wrong step (propose an amendment and continue on unaffected parts), preventing silent workarounds. - Use Case: You receive a multi-step implementation plan from a previous session. This Skill loads it, executes each step on a branch, requests code review after interface changes and at the end, and reports exactly what was completed, deviated, or left out. ## Quick Start Execute the implementation plan in docs/plan.md step by step, verifying each task and reporting any deviations.

Frequently Asked Questions about executing-plans

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

FAQPage Schema
How do I execute an implementation plan step by step with an AI assistant?▼

Load the entire plan first and read it critically for contradictions or stale assumptions before writing code. Then create a task per plan item, follow each step as written, run its specified verification, and only mark it complete after the check passes.

What should I do when a plan step is blocked during execution?▼

Stop and report what blocked you and what you tried, rather than forcing through with an unsanctioned workaround. If the step is clear but now wrong given earlier findings, propose an amendment and continue on the parts it doesn't touch.

When should code review happen during plan execution?▼

Request review after any task that changes a public interface, after the final task before finishing the branch, and whenever you deviated from the plan. Reviewing after every small task costs more attention than it returns.

Can a plan be executed by parallel subagents instead of sequentially?▼

Sequential self-execution is the default. Fanning out to subagents requires the user to explicitly ask for it, since parallel execution changes coordination costs and review dynamics.

Why is marking tasks complete without running verification a problem?▼

A completion claim backed by an unrun check hides failures until they surface downstream. If a step lacks a specified verification, add one and disclose that you did, rather than assuming it passes.