A structured workflow framework for AI-powered software development using specialized agent roles.
This project defines a systematic approach to software development that leverages specialized AI "subagents" working in sequence to ensure quality, consistency, and thorough planning throughout the development lifecycle.
The framework enforces a disciplined workflow where different AI agents handle specific aspects of development - from initial project understanding through final approval - creating a more structured and reliable development process.
The framework defines eight distinct roles that must be followed in sequence:
- 🎯 Onboarding Specialist - Analyzes new projects and tasks, provides initial understanding
- 📋 Product Owner - Ensures clear project requirements and task understanding
- 📊 Product Manager - Creates detailed implementation plans and technical approaches
- ⚙️ Staff Engineer - Implements according to the plan, requests clarification as needed
- 🧪 QA Engineer - Tests and validates all work against requirements
- 📊 Product Manager - Reviews completed work and provides feedback
- 📋 Product Owner - Provides final approval of the deliverable
- ✅ Task Completion - Work is considered complete
- Plan-First Approach: All engineering work requires detailed planning before implementation
- Quality Gates: Multiple validation points throughout the development lifecycle
- Role Separation: Clear responsibilities for each phase of development
- Documentation-Driven: Comprehensive documentation requirements at every step
- Reflection Mechanism: "What would John Carmack do?" decision point for plan deviations
- Boundary Enforcement: Clear definitions of what changes and what stays unchanged
- Validation Points: Specific checkpoints for quality assurance
- Structured Planning: Comprehensive template for implementation plans
- Intent Mapping: Each step maps to specific project intents and boundaries
- Progress Tracking: Built-in progress tracking with markdown checkboxes
- Rollback Plans: Pre-defined recovery procedures for each implementation
claude-code-subagents/
├── README.md # This file
├── CLAUDE.md # Claude Code integration guide
└── docs/
├── PRD.md # Product requirements (placeholder)
├── START.md # Core workflow guidelines
└── implementation-plan/
└── TEMPLATE.md # Implementation plan template
- AI development tool that supports subagent workflows (e.g., Claude Code)
- Git for version control
- Text editor for documentation
- Start with Onboarding: Always begin any task by having an onboarding specialist analyze the project and task
- Get Product Owner Clarity: Ensure the product owner provides clear project understanding
- Create Implementation Plan: Have the product manager create a detailed plan using the template
- Follow the Plan: Staff engineer implements according to the plan, requesting clarification as needed
- Validate Work: QA engineer tests and validates against requirements
- Review and Approve: Product manager reviews, then product owner approves
- Use the implementation plan template in
docs/implementation-plan/TEMPLATE.md - Record lessons learned in a
docs/memory.mdfile - Reference Context 7 MCP for accessing latest documentation
- Follow the guidelines in
docs/START.md
The framework includes a comprehensive template that covers:
- Intent Mapping: Each step maps to project intents with clear boundaries
- Technical Approach: Specific tools and validation points for each step
- Test Strategy: Testing approach, cases, and coverage requirements
- Success Criteria: Measurable outcomes and quality gates
- Rollback Plans: Recovery procedures and validation steps
- Progress Tracking: Markdown checkboxes for task completion
- Feedback Mechanisms: Structured way for executors to request help
- ✅ Complete workflow definition and guidelines
- ✅ Implementation plan template
- ✅ Integration guide for Claude Code
- ✅ Structured development process
- 🔄 Reference implementation of the subagent system
- 🔄 Integration examples with popular development tools
- 🔄 Metrics and measurement capabilities
- 🔄 Community feedback integration
We welcome contributions to help develop this framework! Areas where we need help:
- Clarify the Context 7 MCP integration requirements
- Expand on the John Carmack decision framework
- Add real-world usage examples
- Build reference implementation of the subagent workflow
- Create integrations with popular development tools
- Develop metrics and measurement systems
- Create test cases for the workflow process
- Validate the framework with real development projects
- Gather feedback from development teams
This framework is built on several key principles:
- Structure Over Speed: Prioritizes thorough planning and validation over rapid iteration
- Documentation First: Heavy emphasis on clear documentation and requirements
- Quality Gates: Multiple validation points prevent issues from propagating
- Role Clarity: Clear separation of concerns across different development phases
- Reflection: Built-in mechanisms for questioning and validating decisions
This framework is designed for:
- Complex Software Projects: Where thorough planning and validation are critical
- Team Development: Where role clarity and handoffs are important
- Quality-Critical Systems: Where multiple validation points add value
- Learning and Training: Where structured processes help build good habits
This framework represents an experiment in structured AI-powered development. Key questions we're exploring:
- How much process structure is optimal for different project types?
- What's the right balance between thorough planning and agile iteration?
- How can AI agents best collaborate on complex development tasks?
- What measurements best indicate the success of this approach?
[Add your chosen license here]
- Create issues for bugs or feature requests
- Start discussions for questions about the framework
- Check existing documentation in the
docs/directory
This project is an exploration of structured AI-powered development workflows. We're actively seeking feedback and contributions from the development community.
Built with ❤️ by BLEN.
BLEN, Inc is a digital services company that provides Emerging Technology (ML/AI, RPA), Digital Modernization (Legacy to Cloud) and Human-Centered Web/Mobile Design and Development.