moai-foundation-core（摩艾基础核心）

MoAI Foundation Core

Foundational principles and architectural patterns that power MoAI-ADK's AI-driven development workflow.

Core Philosophy: Quality-first, test-driven, modular, and efficient AI development through proven patterns and automated workflows.

Quick Reference (30 seconds)

What is MoAI Foundation Core? Six essential principles that ensure quality, efficiency, and scalability in AI-powered development:

1. TRUST 5 Framework - Quality gate system (Test-first, Readable, Unified, Secured, Trackable)
2. SPEC-First TDD - Specification-driven test-driven development workflow
3. Delegation Patterns - Task orchestration via specialized agents (never direct execution)
4. Token Optimization - 200K budget management and context efficiency
5. Progressive Disclosure - Three-tier knowledge delivery (Quick → Implementation → Advanced)
6. Modular System - File splitting and reference architecture for scalability

Quick Access:

• Quality standards → TRUST 5 Module
• Development workflow → SPEC-First TDD Module
• Agent coordination → Delegation Patterns Module
• Budget management → Token Optimization Module
• Content structure → Progressive Disclosure Module
• File organization → Modular System Module
• Agent catalog → Agents Reference Module NEW
• Command reference → Commands Reference Module NEW
• Security & constraints → Execution Rules Module NEW

Use Cases:

• New agent creation with quality standards
• New skill development with structural guidelines
• Complex workflow orchestration
• Token budget planning and optimization
• Documentation architecture design
• Quality gate configuration

Implementation Guide (5 minutes)

1. TRUST 5 Framework - Quality Assurance System

Purpose: Automated quality gates ensuring code quality, security, and maintainability.

Five Pillars:

Integration Points:

• Pre-commit hooks → Automated validation
• CI/CD pipelines → Quality gate enforcement
• Agent workflows → core-quality validation
• Documentation → Quality metrics

Detailed Reference: TRUST 5 Framework Module

2. SPEC-First TDD - Development Workflow

Purpose: Specification-driven development ensuring clear requirements before implementation.

Three-Phase Workflow:

Phase 1: SPEC (/moai:1-plan)
 workflow-spec → EARS format
 Output: .moai/specs/SPEC-XXX/spec.md
 Execute /clear (saves 45-50K tokens)

Phase 2: TDD (/moai:2-run)
 RED: Failing tests
 GREEN: Passing code
 REFACTOR: Optimize
 Validate: ≥85% coverage

Phase 3: Docs (/moai:3-sync)
 API documentation
 Architecture diagrams
 Project reports

EARS Format:

• Ubiquitous: System-wide (always active)
• Event-driven: Trigger-based (when X, do Y)
• State-driven: Conditional (while X, do Y)
• Unwanted: Prohibited (shall not do X)
• Optional: Nice-to-have (where possible, do X)

Token Budget: SPEC 30K | TDD 180K | Docs 40K | Total 250K

Key Practice: Execute /clear after Phase 1 to initialize context.

Detailed Reference: SPEC-First TDD Module

3. Delegation Patterns - Agent Orchestration

Purpose: Task delegation to specialized agents, avoiding direct execution.

Core Principle: Alfred NEVER executes directly. All work via Task().

Delegation Syntax:

result = await Task(
 subagent_type="specialized_agent",
 prompt="Clear, specific task",
 context={"relevant": "data"}
)

Three Patterns:

Sequential (dependencies):

design = Task(subagent_type="api-designer", prompt="Design API")
code = Task(subagent_type="backend-expert", prompt="Implement", context={"design": design})

Parallel (independent):

results = await Promise.all([
 Task(subagent_type="backend-expert", prompt="Backend"),
 Task(subagent_type="frontend-expert", prompt="Frontend")
])

Conditional (analysis-based):

analysis = Task(subagent_type="debug-helper", prompt="Analyze")
if analysis.type == "security":
 Task(subagent_type="security-expert", prompt="Fix")

Agent Selection:

• Simple (1 file): 1-2 agents sequential
• Medium (3-5 files): 2-3 agents sequential
• Complex (10+ files): 5+ agents mixed

Detailed Reference: Delegation Patterns Module

4. Token Optimization - Budget Management

Purpose: Efficient 200K token budget through strategic context management.

Budget Allocation:

Token Saving Strategies:

1. Phase Separation: /clear between phases

• After /moai:1-plan (saves 45-50K)
• When context > 150K
• After 50+ messages

1. Selective Loading: Load only necessary files
2. Context Optimization: 20-30K tokens target
3. Model Selection: Sonnet (quality) | Haiku (speed/cost)

Monitoring: /context command, track budget, suggest /clear

Cost Savings: Haiku 70% cheaper → 60-70% total savings

Detailed Reference: Token Optimization Module

5. Progressive Disclosure - Content Architecture

Purpose: Three-tier knowledge delivery balancing value with depth.

Three Levels:

SKILL.md Structure (≤500 lines):

## Quick Reference (30s)
## Implementation Guide (5min)
## Advanced Patterns (10+min)
## Works Well With

Module Architecture:

• SKILL.md: Entry point, cross-references
• modules/: Deep dives, unlimited
• examples.md: Working samples
• reference.md: External links

File Splitting (when >500 lines):

SKILL.md (500 lines)
 Quick (80-120)
 Implementation (180-250)
 Advanced (80-140)
 References (10-20)

Overflow → modules/[topic].md

Detailed Reference: Progressive Disclosure Module

6. Modular System - File Organization

Purpose: Scalable file structure enabling unlimited content.

Standard Structure:

.claude/skills/skill-name/
 SKILL.md # Core (≤500 lines)
 modules/ # Extended (unlimited)
 patterns.md
 examples.md # Working samples
 reference.md # External links
 scripts/ # Utilities (optional)
 templates/ # Templates (optional)

File Principles:

1. SKILL.md: ≤500 lines, progressive disclosure, cross-references
2. modules/: Topic-focused, no limits, self-contained
3. examples.md: Copy-paste ready, commented
4. reference.md: API docs, resources

Cross-Reference Syntax:

Details: [Module](modules/patterns.md)
Examples: [Examples](examples.md#auth)
External: [Reference](reference.md#api)

Discovery Flow: SKILL.md → Topic → modules/[topic].md → Deep dive

Detailed Reference: Modular System Module

Advanced Implementation (10+ minutes)

Cross-Module Integration

TRUST 5 + SPEC-First TDD:

spec = Task(subagent_type="workflow-spec", prompt="SPEC with TRUST 5")
impl = Task(subagent_type="workflow-tdd", prompt="≥85% coverage",
 context={"spec": spec, "quality_gates": ["TRUST5"]})
validation = Task(subagent_type="core-quality", prompt="Validate TRUST 5",
 context={"implementation": impl})

Token-Optimized Delegation:

spec = Task(subagent_type="workflow-spec", prompt="Generate SPEC")
execute_clear() # Save 45-50K
results = await Promise.all([
 Task(subagent_type="backend-expert", prompt="Backend", context={"spec_id": spec.id}),
 Task(subagent_type="frontend-expert", prompt="Frontend", context={"spec_id": spec.id})
])
Task(subagent_type="workflow-docs", prompt="Docs", context={"results": results})

Progressive Agent Workflows:

quick = Task(subagent_type="debug-helper", prompt="Quick diagnosis")
if quick.complexity == "high":
 detailed = Task(subagent_type="debug-helper", prompt="Detailed analysis")
 if detailed.requires_expert:
 expert = Task(subagent_type="security-expert", prompt="Deep dive")

Quality Validation

Pre-Execution:

def validate_execution_requirements(task, context):
 return all([
 validate_security_clearance(task),
 validate_resource_availability(context),
 validate_quality_standards(task),
 validate_permission_compliance(task)
 ])

Post-Execution:

def validate_execution_results(result, task):
 validations = [
 validate_output_quality(result),
 validate_security_compliance(result),
 validate_test_coverage(result),
 validate_documentation_completeness(result)
 ]
 if not all(validations):
 raise QualityGateError("Quality gate failures")
 return True

Error Handling

Delegation Failure:

try:
 result = Task(subagent_type="backend-expert", prompt="Complex task")
except AgentExecutionError as e:
 analysis = Task(subagent_type="debug-helper", prompt=f"Analyze: {e}")
 if analysis.issue == "complexity":
 results = await Promise.all([
 Task(subagent_type="backend-expert", prompt="Subtask 1"),
 Task(subagent_type="backend-expert", prompt="Subtask 2")
 ])

Token Budget Exceeded:

if token_usage > 150_000:
 execute_clear()
 context = {"spec_id": current_spec.id, "phase_results": summarize(previous_results)}
 Task(subagent_type="next-agent", prompt="Continue", context=context)

Works Well With

Agents:

• agent-factory - Create agents with foundation principles
• skill-factory - Generate skills with modular architecture
• core-quality - Automated TRUST 5 validation
• workflow-spec - EARS format specification
• workflow-tdd - RED-GREEN-REFACTOR execution
• workflow-docs - Documentation with progressive disclosure

Skills:

• moai-cc-claude-md - CLAUDE.md with foundation patterns
• moai-cc-configuration - Config with TRUST 5
• moai-cc-memory - Token optimization
• moai-core-ask-user-questions - User clarification
• moai-context7-integration - MCP integration

Commands:

• /moai:1-plan - SPEC-First Phase 1
• /moai:2-run - TDD Phase 2
• /moai:3-sync - Documentation Phase 3
• /moai:9-feedback - Continuous improvement
• /clear - Token management

Foundation Modules (Extended Documentation):

• Agents Reference - 26-agent catalog with 7-tier hierarchy
• Commands Reference - 6 core commands workflow
• Execution Rules - Security, Git strategy, compliance

Quick Decision Matrix

Module Deep Dives:

• TRUST 5 Framework
• SPEC-First TDD
• Delegation Patterns
• Token Optimization
• Progressive Disclosure
• Modular System
• Agents Reference NEW
• Commands Reference NEW
• Execution Rules NEW

Full Examples: examples.md External Resources: reference.md

Version: 2.2.0 Last Updated: 2025-11-26 Status: Active (496 lines, within 500-line limit)