Skills writing-plans
📋

writing-plans

Safe 🌐 Network access⚙️ External commands

Create implementation plans for features

Also available from: Cycleaddict,obra,DYAI2025,DMJGilbert,davila7,CodingCossack

Breaking complex features into actionable tasks is difficult and error-prone. This skill provides a structured template for writing comprehensive implementation plans with test-driven steps, exact file paths, and verification commands.

Supports: Claude Codex Code(CC)
📊 70 Adequate
1

Download the skill ZIP

2

Upload in Claude

Go to Settings → Capabilities → Skills → Upload skill

3

Toggle on and start using

Test it

Using "writing-plans". Create an implementation plan for adding a caching layer to the API responses

Expected outcome:

  • Implementation plan saved to docs/plans/caching-layer.md
  • 6 bite-sized tasks with TDD approach
  • Each task includes file paths, test code, implementation code, verification commands
  • Plan header with goal, architecture, and tech stack
  • Ready for subagent-driven or parallel execution

Using "writing-plans". Write a plan to add dark mode theming to the React application

Expected outcome:

  • Implementation plan saved to docs/plans/dark-mode.md
  • 5 tasks covering theme context, components, and CSS variables
  • Test-first approach with failing test steps
  • Exact component file paths and CSS module references
  • Verification commands for visual regression testing

Using "writing-plans". Create an implementation plan for a new REST API endpoint

Expected outcome:

  • Implementation plan saved to docs/plans/new-endpoint.md
  • 4 tasks for model, controller, route, and integration tests
  • Complete test code examples for each task
  • Git commit commands with conventional commit format
  • Ready for subagent dispatch with executing-plans skill

Security Audit

Safe
v5 • 1/17/2026

This is a pure markdown documentation skill containing only planning templates and example code blocks. The previous audit correctly identified no executable code, file access, network calls, or command execution capability. All 29 static findings are false positives caused by markdown syntax and content hash strings being misinterpreted as security vulnerabilities.

2
Files scanned
294
Lines analyzed
2
findings
5
Total audits
Audited by: claude View Audit History →

Quality Score

38
Architecture
100
Maintainability
87
Content
22
Community
100
Security
91
Spec Compliance

What You Can Build

Feature roadmap planning

Transform feature specs into detailed implementation guides that junior developers can follow independently

Task breakdown documentation

Create standardized implementation plans with clear file references and verification steps

Structured task execution

Generate plans that enable systematic subagent-driven development with code review checkpoints

Try These Prompts

Simple feature plan
Create an implementation plan for adding user authentication. Use the writing-plans skill. Save to docs/plans/auth-system.md
API endpoint plan
Write a plan for creating a new REST API endpoint at /api/users/{id}/preferences. Include model, controller, route, and tests.
Refactoring plan
Create a plan to refactor the legacy payment module. Break into small tasks following TDD with test-first approach.
Complex feature plan
Plan implementation for adding real-time notifications across the platform. Include WebSocket setup, notification service, and frontend components. Use subagent-driven execution.

Best Practices

  • Always include exact file paths for every file touched in the plan
  • Provide complete code examples, never partial snippets or comments like 'add validation'
  • Include exact commands with expected output for verification steps

Avoid

  • Vague task descriptions like 'implement feature' without specific actions
  • Skipping test-first steps or providing incomplete test code
  • Not specifying exact file paths or using relative paths without context

Frequently Asked Questions

Which AI tools support this skill?
This skill works with Claude and Claude Code. It is designed for Claude Code subagent execution workflows.
What is the ideal task granularity?
Each task should take 2-5 minutes. Include write test, run test, write code, run test, commit.
How do I integrate with other skills?
Use @ syntax to reference relevant skills. Assumes brainstorming skill created a worktree first.
Is my codebase data safe?
Yes. This is a prompt-only skill. It generates markdown plans without executing code or accessing files.
Why is my plan not saving?
Ensure the docs/plans directory exists. The skill requires a worktree created by the brainstorming skill.
How is this different from regular prompting?
This skill enforces a strict template with test-first steps, exact file paths, and verification commands.

Developer Details

File structure

📄 SKILL.md