Iterative project development system for Claude Code based on Rational Unified Process (RUP)
Rumiator helps you build software projects systematically by combining specialized AI agents with structured workflows.
# 1. Initialize your project
/rumiator-init
# 2. Create product vision
/rumiator-create-product
# 3. Generate tasks (with business requirements)
/rumiator-create-tasks
# 4. Analyze technical approach
/rumiator-analyze-tech all
# 5. Start developing
/rumiator-develop-next
# 6. Monitor progress
/rumiator-statusβ‘ New: Business requirements now integrated into task creation - faster workflow! π¨ New: Customize commands and agents to fit your workflow - without losing changes on updates! π New: Automatic release system - updates use stable releases instead of master branch!
- β Systematic workflow - From idea to implementation
- β Specialized agents - PM, Analyst, Architect, Developers, DevOps, QA
- β Comprehensive documentation - Auto-generated specs and ADRs
- β Iterative development - Small, manageable iterations
- β Progress tracking - Visual dashboards and reports
- β Decision transparency - All choices documented
- β Bug tracking & management - Track bugs alongside features
- β Architecture evolution - Review and update architectural decisions
- β Customizable workflows - Override commands and agents to match your team's needs
- β Automatic releases - GitHub Actions automatically creates releases when changelog is updated
- β
Stable updates -
/rumiator-updateuses published releases instead of master branch - β You're in control - Agents ask before important decisions
- project-manager - Product vision and iteration planning
- functional-analyst - Creates tasks with business requirements (summary, user stories, acceptance criteria)
- architect - High-level technical guidance and architectural decisions (ADRs)
- developer-frontend - Frontend implementation based on requirements and guidance
- developer-backend - Backend implementation based on requirements and guidance
- devops - CI/CD and infrastructure
- quality-assurance - Testing and validation
Planning & Setup
| Command | Purpose |
|---|---|
/rumiator-init |
Initialize project structure |
/rumiator-create-product |
Create product plan from idea |
/rumiator-update-plan |
Update product plan |
/rumiator-create-tasks |
Generate tasks from plan |
Analysis & Design
| Command | Purpose |
|---|---|
/rumiator-analyze-business |
|
/rumiator-analyze-tech |
Create high-level technical guidance |
/rumiator-adr |
Create architecture decision record |
Development
| Command | Purpose |
|---|---|
/rumiator-develop |
Implement a task (feature, bug, or review) |
/rumiator-develop-next |
Auto-select and develop next task |
Bug Management
| Command | Purpose |
|---|---|
/rumiator-create-bug |
Create a bug report |
/rumiator-triage-bugs |
Review and prioritize bugs |
Architecture Management
| Command | Purpose |
|---|---|
/rumiator-review-architecture |
Challenge and update architectural decisions |
/rumiator-propagate-architecture-change |
Update tasks affected by architecture changes |
Monitoring & Reporting
| Command | Purpose |
|---|---|
/rumiator-status |
Show project dashboard (features, bugs, reviews) |
/rumiator-report |
Generate iteration report |
/rumiator-block-task |
Mark task as blocked |
Rumiator Tool
| Command | Purpose |
|---|---|
/rumiator-update |
Update to latest Rumiator version |
/rumiator-add-customization |
Create agent or command customizations interactively |
your-project/
βββ .rumiator/
β βββ config.yml # Configuration
β βββ templates/ # Document templates
βββ docs/
β βββ product/ # Product plan, architecture
β βββ features/ # Technical specs (high-level guidance)
β β βββ [feature]/
β β βββ technical.md # β‘ Simplified: architectural guidance only
β β βββ bugs/ # Bug analysis documents
β βββ adr/ # Architecture decisions
β β βββ reviews/ # Architecture review documents
β βββ iterations/ # Iteration reports & plans
β βββ iteration-01/ # Current iteration
β β βββ tasks/ # Task definitions (features, bugs, reviews)
β β β βββ TASK-001.yml # type: feature
β β β βββ TASK-015.yml # type: bug
β β β βββ TASK-040.yml # type: architecture-review
β β βββ plan.md
β β βββ report.md
β β βββ bug-triage-report.md
β β βββ architecture-change-*.md
β βββ iteration-02/ # Future iterations
β βββ ...
βββ repositories/ # Source code
# Initialize
/rumiator-init
# Describe your product
/rumiator-create-product
> "A task management app for remote teams..."
# Generate tasks for iteration 1 (includes business requirements)
/rumiator-create-tasks
# Created: TASK-001 (Auth), TASK-002 (Tasks CRUD), TASK-003 (Dashboard)
# Each task includes: summary, user stories, acceptance criteria
# Get technical guidance
/rumiator-analyze-tech all
# Agent asks: "Should I create an ADR for authentication approach?"
> "Yes, use JWT"
# Agent creates high-level technical guidance (no code/schemas)
# Develop tasks
/rumiator-develop-next
# Implements TASK-001 (backend + frontend + tests)
# Developer agents make implementation decisions
# Check progress
/rumiator-status
# Progress: ββββββββββ 33% (1/3 tasks done)
# Continue development
/rumiator-develop-next
# Repeat until iteration complete
# Generate report
/rumiator-report
# Creates docs/iterations/iteration-01/report.mdβ‘ Simplified: 2-phase workflow instead of 3 - faster and more efficient!
- Iterative - Deliver value in 2-4 week iterations
- Documented - Every decision is recorded
- Interactive - Agents ask when uncertain
- Transparent - Track progress visually
- Quality-focused - Specs, tests, reviews built-in
π Complete Guide - Comprehensive documentation
Topics covered:
- Detailed workflow explanation
- Agent responsibilities
- Task lifecycle
- Best practices
- Customization guide
- FAQ
Rumiator uses an automated release system to ensure stable updates:
When a new version is documented in RUMIATOR_CHANGELOG.md, GitHub Actions automatically:
- β
Detects the new version (format:
[X.Y.Z] - YYYY-MM-DD) - β Compares with the latest published release
- β
Ensures
config.yml.templatehas matchingrumiator_version - β Creates a GitHub Release with changelog summary and comparison link
When you run /rumiator-update, it:
- β Uses the latest published release (not master branch)
- β Ensures you get tested, stable versions
- β Falls back to master only if releases are unavailable (with warning)
- Stability - Updates use verified releases
- Traceability - Each release has clear changelog
- Consistency - Version always matches between changelog and config
- Automation - No manual release creation needed
- Claude Code (with agent and command support)
- Git repository (recommended)
Rumiator uses semantic versioning and supports automatic updates:
Your Rumiator version is stored in .rumiator/config.yml under rumiator_version
/rumiator-updateThis command will:
- Clone/update the official repository to
repositories/claude-rumiator/ - Compare your version with the latest available
- Apply necessary migrations automatically
- Update your project structure and configuration
See RUMIATOR_CHANGELOG.md for detailed version history and migration instructions.
If automatic update fails, you can:
- Check
repositories/claude-rumiator/RUMIATOR_CHANGELOG.mdfor migration steps - Manually update your
.rumiator/directory - Update
rumiator_versionin yourconfig.yml
Use the /rumiator-workflow-pr command to create a release:
/rumiator-workflow-prThis command will:
- Analyze all changes since the last release
- Automatically detect new commands, templates, and config changes
- Generate migration actions based on detected changes
- Ask you a few questions (version type, summary, etc.)
- Update RUMIATOR_CHANGELOG.md and config.yml.template
- Commit changes and create a PR
Then:
- Review and merge the PR
- Create a git tag:
git tag -a v1.x.x -m "Release v1.x.x" && git push origin v1.x.x - Create a GitHub Release from the tag
Much easier than writing CHANGELOG manually! π
- Structured approach to building projects
- Comprehensive documentation without extra effort
- Clear progress tracking
- Reduces decision fatigue
- Shared understanding of requirements
- Clear technical specifications
- Documented architectural decisions
- Easy onboarding for new members
- See how professional projects are structured
- Learn RUP methodology in practice
- Understand architecture decision-making
- Study code organization patterns
All agents and workflows are customizable:
- Edit
.claude/agents/*.mdto change agent behavior - Edit
.claude/commands/*.mdto modify workflows - Edit
.rumiator/templates/*.mdto adjust document structure
π‘ Run /rumiator-status frequently to track progress
π‘ Review generated docs and refine them
π‘ Let agents ask questions - it improves quality
π‘ Keep iterations small (5-15 tasks)
π‘ Commit after each completed task
After completing an iteration:
- Review with
/rumiator-report - Update plan if needed with
/rumiator-update-plan - Plan next iteration
- Repeat the cycle
Improve Rumiator by:
- Enhancing agent prompts
- Adding new workflows
- Creating new templates
- Sharing your improvements
MIT - Use freely for any project
Get Started: Run /rumiator-init in your Claude Code session! π
For questions and detailed documentation, see RUMIATOR-GUIDE.md