Claude Code Plugins

Community-maintained marketplace

Feedback

Execute implementation plans step-by-step with quality guardrails (ask when unclear, minimal-change bias, readability checklist, verification). Use when the user asks to execute or follow a plan, run tasks sequentially from a plan file (e.g., docs/plans/*.md), or continue plan steps.

Install Skill

1Download skill
2Enable skills in Claude

Open claude.ai/settings/capabilities and find the "Skills" section

3Upload to Claude

Click "Upload skill" and select the downloaded ZIP file

Note: Please verify skill by going through its instructions before using it.

SKILL.md

name executing-plans
description Execute implementation plans step-by-step with quality guardrails (ask when unclear, minimal-change bias, readability checklist, verification). Use when the user asks to execute or follow a plan, run tasks sequentially from a plan file (e.g., docs/plans/*.md), or continue plan steps.

Executing Plans

Overview

Execute an existing plan task-by-task with strong clarity, minimal change, and verification. The plan is the source of truth; do not improvise or refactor beyond it without asking.

Quick Start

  • Announce: "I'm using the executing-plans skill to run the plan."
  • Locate plan: Use the provided file path or inline plan text; ask if missing.
  • Confirm start: Ask which task to begin with if not explicit.

Execution Workflow

Pre-flight

  • Read the target plan section.
  • Identify any unclear items (missing files, ambiguous commands, unclear intent).
  • Ask a single, precise question if anything is unclear. Do not guess.

Task Loop (for each Task N)

  • Restate the step goal in plain language.
  • Open/inspect only the files listed in the task.
  • Follow the plan steps in order (test → run → implement → run → checklist).
  • Do not commit. If the plan includes commit steps, skip them and note the skip.
  • Use minimal changes to satisfy the step; avoid refactors unless explicitly required.
  • If a new file or modification seems necessary but not listed, ask first.

Ambiguity Rule

  • Ask whenever something is unclear, regardless of the step.
  • Ask one question at a time with concrete options when possible.

Minimal-Change Bias

  • Preserve existing style, structure, and formatting.
  • Touch only the files listed in the task.
  • Avoid reformatting or cleanup unless the plan says so.

Readability/Style Checklist (after each task or file edit)

  • Naming: Clear, explicit names; no ambiguity.
  • Structure: Minimal and localized; no unnecessary nesting or indirection.
  • Readability: Intent is obvious to a new reader.
  • Errors: Failure modes are explicit; no silent failure unless intended.
  • Consistency: Matches existing patterns in the file.
  • Docs/Comments: Add only if intent is not obvious from code.

If the checklist suggests changes beyond the plan, ask before doing them.

Verification and Failures

  • Run the exact commands specified in the plan when possible.
  • Do not claim completion without verification.
  • If tests fail or behavior is unexpected, switch to the systematic-debugging skill (if available).
  • If verification is blocked (permissions or environment), explain and ask for next steps.

Reporting and Progress

  • After each task: summarize changes with file paths and tests run (plus results).
  • Ask to proceed to the next task.