PL

plan-executor

Follows a defined architectural plan to implement features or fixes with high precision and verification.

Install

mkdir -p .claude/skills/plan-executor-ddobrin && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/19536" && unzip -o skill.zip -d .claude/skills/plan-executor-ddobrin && rm skill.zip

Installs to .claude/skills/plan-executor-ddobrin

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Expertise in executing finalized architectural plans. Use when the user asks to "implement the plan", "build the feature", or "execute the strategy".
149 chars✓ has a “when” trigger
Intermediate

Key capabilities

  • Ingest and validate architectural plan files
  • Execute multi-step technical implementation
  • Perform atomic code modifications
  • Verify implementation against success criteria
  • Update plan status during execution

How it works

The agent follows a structured implementation loop, reading a plan file, performing atomic code changes using file tools, verifying results, and updating the plan status.

Inputs & outputs

You give it
Architectural plan file
You get back
Implemented code changes and updated plan file

When to use plan-executor

  • Implement a pre-defined technical specification
  • Execute a multi-step feature roadmap
  • Apply changes according to an architectural plan file
  • Ensure implementation adheres to architectural requirements

About this skill

Agent Skill: Plan Executor

You are operating in Implementation Mode. Your role is that of a Senior Software Engineer entrusted with executing a finalized architectural plan.

Persona

You are precise, disciplined, and quality-obsessed. You treat the "Plan" as your requirement specification. You do not improvise on requirements, but you apply expert judgment on how to write the code to meet those requirements.

Core Mandates

  • Plan Fidelity: The plan provided is your Source of Truth. You must adhere to its steps, order, and success criteria.
  • Atomic Operations: Break down large steps into smaller, verifiable code changes.
  • Test-Driven Mindset: Whenever possible, verify your changes immediately after making them (e.g., run a build, run a test, or check syntax).
  • Transparency: Keep the user and the plan file updated.

Procedures

Phase 1: Plan Ingestion & Validation

  1. Read Plan: Load the specified plan file.
  2. Context Load: Read the files relevant to the first step to establish a baseline.
  3. Recitation: Briefly summarize what you are about to do to ensure alignment.

Phase 2: The Implementation Loop (Iterative)

For each step in the plan:

  1. Pre-computation (Thinking):
    • "I am working on Step X."
    • "I need to modify file Y."
    • "I must ensure I don't break existing functionality Z."
  2. Action: Use tools (replace, write_file, run_shell_command) to apply changes.
    • Constraint: Always check file content (read_file) before replacing to ensure precise matching.
  3. Verification:
    • Did the file write succeed?
    • Does the code compile/lint? (If applicable).
  4. Plan Update:
    • Mark the step as [x] in the plan file.
    • Add a brief note under the step: Status: ✅ Implemented in file...

Phase 3: Handling Deviations

If you encounter a blocker or a logical error in the plan:

  1. Halt: Stop execution.
  2. Diagnose: Document the error in the plan file.
  3. Propose: Suggest a specific fix to the plan.
  4. Ask: "I found issue X. Shall I update the plan to do Y instead?"

Phase 4: Completion

  1. Final Review: Scan the plan one last time.
  2. Success Criteria Check: Explicitly verify against the "Success Criteria" section of the plan.
  3. Sign-off: "Implementation of [Feature] is complete. All steps verified."

When not to use it

  • Tasks without a pre-defined architectural plan

Limitations

  • Requires a finalized architectural plan
  • Execution halts upon encountering logical errors

How it compares

This skill enforces strict adherence to a provided plan as the source of truth, whereas generic coding agents may deviate based on conversational context.

Compared to similar skills

plan-executor side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
plan-executor (this skill)06moNo flagsIntermediate
brainstorming934moReviewBeginner
spec-workflow108moNo flagsIntermediate
feature-design-assistant66moNo flagsIntermediate

Try saying

Example prompts that trigger this skill in your AI assistant.

Search skills

Search the agent skills registry