An ultra-intelligent system that analyzes project needs and dynamically creates specialized subagents for parallel development. Detects both current requirements and anticipates future needs based on project structure, dependencies, code patterns, and development trajectories.
The Agent Management System enables you to:
- Analyze Projects: Automatically detect technology stacks, patterns, and needs
- Create Specialized Agents: Generate subagents with specific expertise
- Parallel Development: Launch multiple agents working simultaneously
- Predictive Intelligence: Anticipate future project requirements
/Users/besi/.claude/agents/
├── agent-manager.md # Main orchestrator agent
├── agent-creator.js # Agent creation workflow
├── agent-templates.js # 10+ specialized agent templates
├── project-analyzer.js # Project analysis engine
├── predictive-analyzer.js # Future needs prediction
├── agent-manager-cli.js # Command-line interface
└── agent-manager # Bash wrapper script
- frontend-architect: React/Vue/Angular expert, UI/UX implementation
- backend-engineer: API design, database optimization, server logic
- fullstack-developer: End-to-end feature development and integration
- mobile-developer: React Native, Flutter, native iOS/Android
- devops-engineer: CI/CD, infrastructure, deployment automation
- security-specialist: Security audits, vulnerability assessments
- performance-optimizer: Performance analysis, optimization, monitoring
- testing-expert: Test strategy, automation, quality assurance
- api-designer: REST/GraphQL/gRPC design, documentation
- database-architect: Schema design, optimization, migrations
- cloud-specialist: AWS/GCP/Azure, cloud-native architecture
- ai-ml-engineer: Machine learning, data processing, model deployment
- blockchain-developer: Smart contracts, DApps, Web3 integration
- documentation-writer: Technical documentation, API docs
- code-reviewer: Code quality, best practices, standards enforcement
- integration-specialist: Third-party integrations, API connections
- migration-expert: Legacy system modernization, data migration
Use the agent manager to analyze project needs:
@agents/agent-manager.md
The agent manager will:
- Scan project structure and files
- Detect technology stack and frameworks
- Identify current needs and gaps
- Recommend specialized agents
Let the agent manager create specialized subagents:
@agents/agent-manager.md
Create agents for [your-project-path]
The agent manager will:
- Create
.mdagent files in/project-folder/.claude/agents/ - Generate agents based on detected needs
- Set up coordination protocols
Launch created agents for parallel work:
@agents/frontend-architect.md
@agents/backend-engineer.md
@agents/testing-expert.md
# Agent manager creates:
- frontend-architect.md (React/TypeScript specialist)
- testing-expert.md (Jest/React Testing Library)
- performance-optimizer.md (Bundle optimization)
- documentation-writer.md (API docs)# Agent manager creates:
- backend-engineer.md (Server logic)
- database-architect.md (Schema design)
- security-specialist.md (Auth/Security)
- testing-expert.md (API testing)# Agent manager creates:
- frontend-architect.md (UI development)
- backend-engineer.md (API/services)
- devops-engineer.md (CI/CD)
- security-specialist.md (Security)
- testing-expert.md (QA)
- documentation-writer.md (Docs)For any project /project-folder, the system creates:
/project-folder/.claude/
├── agents/
│ ├── frontend-architect.md # Specialized agent
│ ├── backend-engineer.md # Specialized agent
│ ├── testing-expert.md # Specialized agent
│ └── [other-agents].md # Based on needs
└── workflows/
└── team-workflow.md # Coordination guide
Each created agent has full tool access:
- File Operations: Read, Write, Create, Edit, MultiEdit
- Shell Commands: Execute (bash), background processes
- Search: Grep, Glob for code analysis
- Web: WebSearch, FetchUrl for research
- Project Management: TodoWrite for task tracking
- Scan directory structure and files
- Detect technologies (package.json, requirements.txt, etc.)
- Identify patterns and architecture
- Assess complexity and needs
- Current technical challenges
- Skill gaps in team capabilities
- Testing and quality gaps
- Security and performance needs
- Generate specialized agents based on analysis
- Create agent
.mdfiles with proper YAML frontmatter - Configure tools and capabilities
- Set up coordination protocols
- Launch multiple agents simultaneously
- Coordinate through shared TodoWrite
- Prevent conflicts and overlaps
- Track progress and completion
The system anticipates future needs:
- Technology Upgrades: Detects upcoming framework updates
- Scaling Challenges: Predicts performance bottlenecks
- Security Needs: Identifies vulnerability patterns
- Skill Gaps: Anticipates missing capabilities
Located at: /Users/besi/.claude/agents/agent-manager.md
Key capabilities:
- Project analysis and need detection
- Dynamic agent creation
- Coordination setup
- Predictive intelligence
Located at: /Users/besi/.claude/agents/agent-templates.js
Pre-configured templates for:
- All agent types and specializations
- Tool configurations
- Capability definitions
- Focus areas and workflows
- Complex projects requiring multiple expertise areas
- Large codebases with diverse technologies
- Parallel development needed for speed
- Tasks that can be divided by domain/technology
- Projects requiring specialized knowledge
- Use TodoWrite for shared task management
- Coordinate agents through project context
- Prevent overlapping file modifications
- Track progress across all agents
- Start with minimal viable agent set
- Add agents based on actual needs
- Ensure complementary capabilities
- Avoid redundant specializations
The system can create custom agents based on:
- Specific technology stacks
- Unique project requirements
- Team skill combinations
- Domain-specific needs
- Git integration for change tracking
- CI/CD pipeline coordination
- Documentation synchronization
- Quality assurance workflows
- Background task execution
- Parallel processing capabilities
- Resource management
- Progress monitoring
Track effectiveness through:
- Development velocity improvement
- Code quality enhancement
- Bug reduction rates
- Time-to-market acceleration
- Team productivity gains
- Technical debt reduction
- Agent Creation Only: The agent manager ONLY creates agents, doesn't execute project work
- Stop After Creation: Agent manager stops immediately after creating agent files
- Tool Usage: Agents have full tool access including Execute (bash) and Write
- Parallel Execution: Multiple agents can work simultaneously on different tasks
- Coordination Required: Use TodoWrite and shared documentation for coordination
Agents not created?
- Check if agent manager is being used correctly
- Verify project path is accessible
- Ensure agent manager has Create/Write tool permissions
Coordination issues?
- Use TodoWrite for shared task management
- Check agent workflows for coordination protocols
- Verify agent files are in correct directory
Performance problems?
- Monitor agent workload distribution
- Check for overlapping tasks
- Optimize agent assignment based on capabilities
- Agent Manager:
/Users/besi/.claude/agents/agent-manager.md - CLI Interface:
/Users/besi/.claude/agents/agent-manager-cli.js - Templates:
/Users/besi/.claude/agents/agent-templates.js - Project Analysis:
/Users/besi/.claude/agents/project-analyzer.js
The system is designed to be extensible:
- Add new agent templates in
agent-templates.js - Enhance analysis algorithms in
project-analyzer.js - Improve predictive capabilities in
predictive-analyzer.js - Extend CLI functionality in
agent-manager-cli.js
This agent management system is part of the Factory AI development toolkit.
Ready to supercharge your development with intelligent agent orchestration? 🚀
Start with @agents/agent-manager.md to discover your project's potential!