Enterprise Master-Clone Pattern Skill v4.0.0

Skill Metadata

Field	Value
Skill Name	moai-alfred-clone-pattern
Version	4.0.0 Enterprise (2025-11-12)
Allowed tools	Read, Bash, Task
Auto-load	On demand for complex multi-step tasks
Tier	Alfred (Orchestration)
Lines of Content	900+ with 12+ enterprise examples
Progressive Disclosure	3-level (quick-start, patterns, advanced)

---

What It Does

Provides comprehensive guidance for Alfred's Master-Clone pattern - a delegation mechanism where Alfred creates autonomous clones (Task-delegated agents) to handle complex multi-step tasks that don't require domain-specific expertise but benefit from:

• Full project context and codebase understanding
• Parallel processing capabilities
• Independent decision-making
• Comprehensive state tracking

---

When to Use (Decision Framework)

Use Clone Pattern when:

• Task requires 5+ sequential steps OR affects 100+ files
• No domain-specific expertise needed (not UI, Backend, DB, Security, ML)
• Task is complex with high uncertainty
• Parallel processing would be beneficial
• Full project context is required for optimal decisions
• Task can run autonomously without continuous user input

Examples:

• Large-scale migrations (v0.14.0 → v0.15.2 affecting 200 files)
• Refactoring across many files (100+ imports, API changes)
• Parallel exploration/evaluation tasks
• Complex architecture restructuring
• Bulk file transformations with context-aware logic
• Schema migrations affecting multiple services
• Dependency upgrade cascades

DON'T use Clone Pattern when:

• Domain expertise needed (use specialist agents instead)
• Task < 5 steps (direct execution is faster)
• Quick yes/no decision (use AskUserQuestion)
• Single file modification (use tdd-implementer)

---

Key Concepts

Master-Clone Architecture

Master Agent (Alfred)
    ↓ Creates with Task()
Clone Agent #1          Clone Agent #2          Clone Agent #3
(Parallel execution with shared context)
    ↓                       ↓                       ↓
[Exploration]        [Analysis]             [Implementation]
    ↓                       ↓                       ↓
Results aggregation & synthesis by Master
    ↓
User presentation + next steps

Master Responsibilities:

• Analyze task scope and complexity
• Decompose into independent parallel work
• Create clones with appropriate context
• Aggregate and synthesize results
• Present findings to user

Clone Responsibilities:

• Execute assigned sub-task autonomously
• Use full project context for intelligent decisions
• Track state and progress
• Report findings with evidence
• Handle errors gracefully

---

3-Level Architecture

Level 1: Simple Parallel Task

Scenario: Explore multiple implementation approaches in parallel

// Master agent: Create parallel clones for exploration
const clones = [
  Task({
    description: "Explore PostgreSQL implementation for user persistence",
    prompt: "Analyze PostgreSQL libraries, schema design, migration strategy. Provide pros/cons and code examples."
  }),
  Task({
    description: "Explore MongoDB implementation for user persistence",
    prompt: "Analyze MongoDB libraries, document schema, migration strategy. Provide pros/cons and code examples."
  }),
  Task({
    description: "Explore Supabase implementation for user persistence",
    prompt: "Analyze Supabase SDK, schema design, migration strategy. Provide pros/cons and code examples."
  })
];

// Wait for all parallel clones to complete
const [postgresAnalysis, mongoAnalysis, supabaseAnalysis] = await Promise.all(clones);

// Master synthesizes results
const comparison = {
  options: [postgresAnalysis, mongoAnalysis, supabaseAnalysis],
  recommendation: selectBestOption(clones),
  tradeoffs: analyzeTradeoffs(clones)
};

Level 2: Sequential Dependent Tasks

Scenario: Complex migration where later steps depend on earlier analysis

// Step 1: Analyze current state
const analysisResult = await Task({
  description: "Analyze v0.14.0 codebase structure",
  prompt: "Scan project for all imports of 'old-api'. Document usage patterns, edge cases, and dependencies. Provide summary with file-by-file breakdown."
});

// Step 2: Plan migration strategy (depends on analysis)
const planResult = await Task({
  description: "Plan migration strategy from v0.14.0 to v0.15.2",
  prompt: `Using this analysis: ${analysisResult}\n\nCreate a step-by-step migration plan with:\n- Phased approach (phase 1, 2, 3)\n- Risk mitigation\n- Testing strategy\n- Rollback procedure`
});

// Step 3: Execute migration (depends on plan)
const migrationResult = await Task({
  description: "Execute v0.14.0 → v0.15.2 migration",
  prompt: `Using this plan: ${planResult}\n\nExecute the migration:\n- Update imports\n- Modify APIs\n- Update tests\n- Verify compatibility`
});

// Step 4: Validate results (depends on migration)
const validationResult = await Task({
  description: "Validate migration completeness",
  prompt: `Verify migration:\n- All imports updated\n- No breaking changes\n- Tests passing\n- Performance metrics maintained`
});

// Master reports final state
return {
  analysis: analysisResult,
  plan: planResult,
  migration: migrationResult,
  validation: validationResult,
  status: validationResult.passed ? "SUCCESS" : "NEEDS_REVIEW"
};

Level 3: Hybrid Parallel + Sequential (Advanced)

Scenario: Large refactoring with parallel analysis, synchronized implementation

// Phase 1: Parallel analysis clones
const [apiAnalysis, dbAnalysis, authAnalysis] = await Promise.all([
  Task({ description: "Analyze API layer usage...", prompt: "..." }),
  Task({ description: "Analyze DB layer usage...", prompt: "..." }),
  Task({ description: "Analyze Auth layer usage...", prompt: "..." })
]);

// Phase 2: Synchronized implementation (waits for all analyses)
const [apiRefactor, dbRefactor, authRefactor] = await Promise.all([
  Task({
    description: "Refactor API layer",
    prompt: `Based on analysis:\n${apiAnalysis}\n\nRefactor API with:\n- New patterns\n- Tests\n- Documentation`
  }),
  Task({
    description: "Refactor DB layer",
    prompt: `Based on analysis:\n${dbAnalysis}\n\nRefactor DB with:\n- Schema updates\n- Migration scripts\n- Tests`
  }),
  Task({
    description: "Refactor Auth layer",
    prompt: `Based on analysis:\n${authAnalysis}\n\nRefactor Auth with:\n- New strategy\n- Migration\n- Tests`
  })
]);

// Phase 3: Integration validation
const integrationResult = await Task({
  description: "Validate refactored layer integration",
  prompt: `Verify all refactored layers work together:\n- API uses new DB patterns\n- Auth integrates with API\n- No breaking changes\n- All tests passing`
});

return {
  phase1: { apiAnalysis, dbAnalysis, authAnalysis },
  phase2: { apiRefactor, dbRefactor, authRefactor },
  phase3: integrationResult,
  status: "COMPLETE"
};

---

Best Practices

DO

• Define clear sub-tasks: Each clone should have a specific, measurable goal
• Provide full context: Include project structure, existing patterns, constraints
• Use sequential when needed: Depend on earlier results when necessary
• Validate results: Always verify clone outputs before proceeding
• Document findings: Track what each clone discovered
• Handle failures gracefully: Plan for individual clone failures
• Aggregate intelligently: Synthesize parallel results into coherent analysis

DON'T

• Over-parallelize: Creating 20 clones is overkill (use 2-5)
• Under-specify tasks: Vague descriptions lead to mediocre results
• Ignore dependencies: Force sequential when tasks actually depend
• Skip validation: Trust but verify clone outputs
• Lose context: Always include relevant project information
• Create circular dependencies: Avoid Task A waiting on Task B waiting on Task A

---

Implementation Patterns

Pattern 1: Exploration with Synthesis

// Create 3 clones exploring different approaches
const explorations = await Promise.all(approaches.map(approach =>
  Task({
    description: `Explore ${approach.name} approach`,
    prompt: `Research ${approach.name}...\nProvide: pros, cons, code example, learning curve`
  })
));

// Master synthesizes into comparison table
return {
  comparison: createComparisonTable(explorations),
  recommendation: selectBestApproach(explorations),
  decisionRationale: explainDecision(explorations)
};

Pattern 2: Phased Migration

// Phase 1: Analyze
const analysis = await analyzeCurrentState();

// Phase 2: Plan (depends on analysis)
const plan = await planMigration(analysis);

// Phase 3: Implement (depends on plan)
const implementation = await implementMigration(plan);

// Phase 4: Validate (depends on implementation)
const validation = await validateMigration(implementation);

return { analysis, plan, implementation, validation };

Pattern 3: Distributed Refactoring

// Split files into groups, refactor each group in parallel
const fileGroups = splitFilesIntoGroups(files, 5);
const refactorResults = await Promise.all(
  fileGroups.map(group =>
    Task({
      description: `Refactor files: ${group.join(", ")}`,
      prompt: `Refactor these files using new patterns:\n${group.join("\n")}`
    })
  )
);

// Master validates all refactored files work together
return validateIntegration(refactorResults);

---

When NOT to Use (Anti-Patterns)

Scenario	Why	Use Instead
Single file change	Too much overhead	Direct tdd-implementer
2-3 quick steps	Sequential simpler	Direct execution
Domain expertise required	Needs specialist	Specialist agent (security, DB, etc.)
Real-time interaction	Clones run independently	Interactive agent
Simple query	Overkill complexity	Direct lookup

---

Related Skills

• moai-alfred-agent-guide (Agent architecture & delegation)
• moai-alfred-task-decomposition (Breaking down complex tasks)
• moai-essentials-refactor (Refactoring patterns & examples)

---

For detailed API specifications: reference.md
For real-world examples: examples.md
Last Updated: 2025-11-12
Status: Production Ready (Enterprise v4.0.0)