Planning Plugin for Claude Code
A Claude Code plugin that transforms raw ideas, todos, and brainstorms into well-architected, actionable plans with proper design patterns, Agile/SCRUM structure, and TDD practices.
Part of the Dragonscale Marketplace.
Installation
From Dragonscale Marketplace (Recommended)
First, add the Dragonscale marketplace to Claude Code:
claude /plugin marketplace add Dragonscale-Labs/marketplace
Then install the planning plugin:
claude /plugin install dragonscale/plan
Features
- Proactive Discovery - Asks questions to surface requirements you haven't thought of
- Design Patterns - Silently structures work so proper patterns emerge (Repository, Service Layer, Factory, etc.)
- Agile/SCRUM - Organizes plans with Epics, User Stories, Tasks, and Acceptance Criteria
- Test-Driven Development - Enforces TDD for backend (80%+ coverage), pragmatic for frontend (40-60%)
- Test Review Skill - Dedicated skill to review test quality, coverage, and TDD compliance
- Self-Contained Tasks - Every task includes full context for execution in a fresh session
- Folder Architecture - Creates CLAUDE.md files in code folders so future agents inherit architectural context
- Task Management - Creates tickets in Linear or Jira (auto-detects connected MCPs)
Commands
| Command | Description | Example |
|---|
/plan:refine | Structure raw ideas into plans | /plan:refine Add login, fix bug, update docs |
/plan:session | Save current planning work | /plan:session auth-feature |
/plan:tasks | Create tasks in Linear/Jira | /plan:tasks |
/plan:breakdown | Decompose complex task | /plan:breakdown Implement user authentication |
/plan:test | Review test quality and coverage | /plan:test src/services |
/plan:review | Final code review against task criteria | /plan:review |
Auto-Invoked Skills
Planning Assistant
Automatically activates when Claude detects:
- Raw lists of todos or ideas
- Brainstorming sessions
- Unstructured feature requests
- "Help me plan..." type requests
Test Review
Automatically activates when Claude detects:
- Requests to review tests or test quality
- Coverage reports or discussions
- Questions about what/how to test
- Code reviews involving test files
Test-Driven Development
The plugin enforces TDD with layer-appropriate coverage targets:
| Layer | Coverage Target | Required? |
|---|
| Services/Business Logic | 80%+ | Yes |
| Repositories/Data Access | 80%+ | Yes |
| API/Controllers | 70%+ | Yes |
| Utilities/Helpers | 90%+ | Yes |
| Frontend Components | 40-60% | Critical paths only |
| Frontend Styling | 0% | Skip |
TDD Task Structure:
**Test First**:
- [ ] Write tests for expected behavior
- [ ] Include edge cases and error scenarios
**Then Implement**:
- [ ] Code to pass tests
Task Management Setup
Linear
claude mcp add --transport http linear https://mcp.linear.app/mcp
Jira
Follow your organization's Atlassian Rovo MCP setup instructions.
The /plan:tasks command auto-detects which MCPs are connected.
Workflow
- Dump your raw ideas, todos, or brainstorm notes
- Refine with
/plan:refine or let the auto-skill structure them
- Break down complex items with
/plan:breakdown
- Save your session with
/plan:session
- Create tasks in Linear/Jira with
/plan:tasks
Planning Principles
Proactive Discovery
Every command asks 2-4 targeted questions to surface hidden requirements before structuring plans.
Design Patterns Applied Silently
Plans structure work so these patterns emerge naturally:
- Repository Pattern (data access)
- Service Layer (business logic)
- Factory Pattern (object creation)
- Strategy Pattern (interchangeable behaviors)
- Observer Pattern (events)
- Dependency Injection (testability)
Task 0: Folder Architecture
Every plan includes a "Sprint 0" task that:
- Creates organized folder structure (
models/, services/, repositories/, api/)
- Adds
CLAUDE.md to each folder documenting the pattern
- Enables future agents to understand architecture by reading folder docs
Self-Contained Tasks
Every task includes:
- Context - Why this task exists
- Architectural Purpose - What pattern it implements
- Testing Requirements - What tests to write first
- Implementation Guidance - Senior-level approach
- Acceptance Criteria - How to verify completion
- Anti-patterns - Mistakes to avoid
Session Management
All work is automatically saved to organized session files:
