How does subagent delegation work?

For heavy operations like running full test suites or complex search-and-replace, the skill uses the Task tool to delegate to subagents, keeping the main conversation context lean.

Can I resume a plan if I stop midway?

Yes, the skill reads the plan's status checkboxes to identify exactly where it left off and resumes from the first incomplete task.

Does it support automated git commits?

Yes, the skill creates commits with specified messages after successfully verifying each task or batch, maintaining a clean project history.

How does this skill handle errors during execution?

If a task fails verification, the skill stops execution, reports the error clearly, suggests remediation, and waits for user guidance before proceeding.

What do I need to use the Execute Plan skill?

You need a plan file generated in the docs/plans/ directory, typically created using a /write-plan command or following a specific format.

Execute Plan

Name: Execute Plan
Author: martintrojer

bymartintrojer

•

生产力与工作流

Executes structured development plans sequentially through automated implementation, verification, and progress tracking.

Execute Plan is a specialized Claude Code skill designed to bridge the gap between planning and implementation by systematically following roadmaps generated via the /write-plan command. It manages the entire development lifecycle for a specific task—from making code changes and running verification tests to committing progress and delegating complex sub-tasks to agents—ensuring that projects stay on track and maintain high code quality with minimal token overhead. By automating the transition from roadmap to code, it allows developers to focus on high-level architecture while the skill handles the granular execution steps.

主要功能

01Smart Batching: Groups related small edits to optimize context window usage and execution speed.

023 GitHub stars

03Sequential Task Execution: Processes plan items one-by-one to ensure a logical development flow.

04Progress Tracking: Automatically updates plan files with checkboxes and supports resuming from failures.

05Subagent Delegation: Offloads heavy operations like full-suite testing to subagents for token efficiency.

06Automated Verification: Runs tests or verification commands after every change to prevent regressions.

使用场景

01Executing complex feature additions that require atomic commits and frequent test runs.

02Implementing multi-step refactoring projects based on a pre-defined architectural plan.

03Resuming long-running development tasks across different sessions without losing context.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add martintrojer/dotfiles execute-plan

For use in Claude.ai and ChatGPT