📖 MCP Usage & Examples

Status: ✅ Complete (Session 42.2)

Practical guides, code examples, and best practices for using all 37 MCP tools effectively.

📚 Complete Tool Reference

Core Tools (Everyday Use)

mcp_init

🚀 Smart startup dashboard with mode selection

// Initialize MCP session
mcp_init({ task: "implement user authentication" })

// Pre-select mode
mcp_init({ mode: "quick", skipPrompt: true })

Modes:

• quick - Fast context retrieval (~100 tokens)
• orchestrated - Full decision capture with learning
• search - Semantic search focus
• stats - System analytics

preflight

🧠 Get intelligent context (~100 tokens) before any task

// Basic usage
preflight({ task: "add dark mode toggle" })

// Enforce specific session
preflight({ task: "continue auth work", sessionNumber: 42 })

// Ultra-minimal mode
preflight({ task: "quick fix", lite: true })

Best Practice: Call this BEFORE starting any implementation task.

log_decision

📝 Record development decisions for ML learning

log_decision({
  type: "architecture",  // or "implementation", "refactor", "bugfix", "pattern", "convention"
  question: "How should we handle user authentication?",
  chosen: "JWT tokens with refresh mechanism",
  reasoning: "Better for mobile apps and stateless architecture",
  confidence: 0.85,
  options: ["Session cookies", "JWT tokens", "OAuth only"],
  files: ["src/auth/auth.service.ts", "src/auth/jwt.strategy.ts"]
})

Decision Types:

• architecture - High-level design decisions
• implementation - Coding approach choices
• refactor - Code improvement decisions
• bugfix - Bug resolution choices
• pattern - Design pattern selections
• convention - Code style/standard decisions

find_similar

🔍 Semantic search for past decisions

// Find similar decisions
find_similar({ query: "authentication patterns", limit: 5 })

// Filter by type
find_similar({ query: "CDK drag drop", type: "implementation" })

report_outcome

📊 Report success/failure for reinforcement learning

report_outcome({
  decisionId: "dec_abc123",
  wasSuccessful: true,
  rating: 5,  // 1-5
  feedback: "Clean implementation, no issues in production"
})

Note: Enhanced in Session 42.2 with 4-strategy auto-detection (NX cache, dist folder, test results, terminal logs).

get_conventions / add_convention

📋 Manage learned codebase conventions

// Get conventions
get_conventions({ category: "angular" })

// Add new convention
add_convention({
  category: "components",
  rule: "Always use OnPush change detection",
  examples: ["changeDetection: ChangeDetectionStrategy.OnPush"],
  source: "performance-optimization"
})

get_stats

📈 System statistics and agent performance

get_stats()
// Returns: decision counts, success rates, agent activity, memory usage

get_session_info / set_session

📍 Session tracking and management

// Get current session
get_session_info()

// Enforce specific session (v8.2)
set_session({ sessionNumber: 42, subSessionNumber: 42.2 })

Knowledge Graph Tools (Session 42)

The knowledge graph tracks relationships between files, modules, patterns, and decisions.

graph_create_entity

📍 Create entities in the knowledge graph

graph_create_entity({
  name: "auth.service.ts",
  type: "file",  // file, module, pattern, decision, library, agent, convention
  properties: {
    path: "src/auth/auth.service.ts",
    purpose: "Authentication service"
  }
})

graph_create_relation

🔗 Create relationships between entities

graph_create_relation({
  from: "auth.service.ts",
  to: "jwt.strategy.ts",
  type: "DEPENDS_ON",  // USES, DEPENDS_ON, AFFECTS, CREATED_BY, PART_OF, RELATED_TO, IMPLEMENTS, BREAKS
  weight: 0.9
})

graph_query

🕸️ Natural language graph queries

// Query by natural language
graph_query({ query: "what affects sidebar-nav" })

// Query by entity
graph_query({ entityId: "file:auth.service.ts", direction: "incoming" })

// Filter by type
graph_query({ entityType: "module", relationType: "DEPENDS_ON" })

graph_find_impact

💥 Impact analysis for file/module changes

// What will be affected if I change this file?
graph_find_impact({ fileName: "auth.service.ts" })

// Module impact
graph_find_impact({ moduleName: "authentication" })

graph_build_from_decisions / graph_stats

🏗️ Build and monitor the knowledge graph

// Auto-build graph from existing decisions
graph_build_from_decisions()

// Get graph statistics
graph_stats()

Web & Documentation Search (Session 42)

web_search

🌐 Search the web via Tavily API

web_search({
  query: "Angular 20 standalone components best practices",
  searchType: "documentation",  // general, news, documentation
  maxResults: 5,
  recency: "month",  // day, week, month, year
  saveToMemory: true
})

docs_search

📚 Search library documentation

docs_search({
  query: "drag and drop sortable list",
  libraries: ["@angular/cdk", "rxjs"],
  version: "20.0.0",
  maxResults: 5
})

Code Generation Tools (Session 42.2)

space_generate (NEW)

🎨 QuickGenerate-style code generation for space builders

Inspired by gograb-modules QuickGenerate patterns.

// Quick widget generation
space_generate({
  action: "quick",
  type: "button",
  name: "SubmitButton",
  variant: "primary"
})

// Full widget with all options
space_generate({
  action: "widget",
  type: "table",
  name: "ProductTable",
  features: ["sorting", "filtering", "pagination"]
})

// Generate entire module
space_generate({
  action: "module",
  name: "user-management",
  entities: ["User", "Role", "Permission"],
  features: ["crud", "filters", "export"]
})

// Preview before generating
space_generate({
  action: "preview",
  type: "form",
  name: "LoginForm"
})

// List all available widget types
space_generate({ action: "list_types" })

Supported Widget Types:

• button - Action buttons with variants
• table - Data tables with sorting, filtering, pagination
• form - Forms with validation
• card - Card components
• modal - Modal dialogs
• dashboard - Dashboard layouts
• list - List views
• detail - Detail/view pages
• filters - Filter components
• chart - Chart components
• sidebar - Sidebar navigation
• navbar - Navigation bars
• tabs - Tab components
• accordion - Accordion/collapsible panels
• stepper - Step wizards

analyze_requirement

📋 Analyze user requirements for modules

analyze_requirement({
  description: "Create a product catalog with search and filters",
  features: ["search", "filters", "pagination", "export"],
  context: { existingModules: ["auth", "users"] }
})

validate_module

✅ Validate module against editor2 standards

validate_module({
  modulePath: "modules/business/product-catalog",
  strictMode: true
})
// Returns: validation results with auto-fixable issues

Workflow & Session Recovery (Session 42.2)

workflow (NEW)

🔄 Session recovery and task continuity

Addresses "session continuity broken" gap for cross-agent handoff.

// What was I working on? (cross-agent handoff)
workflow({ action: "recover_session" })

// Quick codebase orientation for new sessions
workflow({ action: "discover_project" })

// Retrieve multi-step task progress
workflow({ action: "get_task_state", taskId: "task_abc123" })

// Save task progress for later recovery
workflow({
  action: "save_task_state",
  description: "Implement authentication flow",
  steps: [
    { name: "Create auth service", status: "completed" },
    { name: "Add JWT strategy", status: "in_progress" },
    { name: "Write tests", status: "pending" }
  ]
})

Advanced ML Tools

predict_outcome

🔮 ML-based outcome prediction before making decisions

predict_outcome({
  decision: {
    type: "implementation",
    question: "Use signals or observables for state?",
    chosen: "Signals for local state",
    reasoning: "Simpler API, better performance",
    confidence: 0.8
  }
})
// Returns: predicted success probability based on historical data

mine_patterns

⛏️ Extract patterns from successful decisions

mine_patterns({
  minSuccessRate: 0.7,
  minSupport: 3,
  patternTypes: ["sequential", "contextual", "outcome"],
  limit: 20
})

transfer_intelligence

🔄 Cross-tier learning transfer

// Transfer learnings from development to platform tier
transfer_intelligence({
  sourceTier: "development",
  targetTier: "platform"
})

Diagnostics & Analytics

get_enhanced_stats

📊 Comprehensive system statistics

get_enhanced_stats()
// Returns: context analysis, git integration, outcome tracking,
// anti-pattern detection, decision capture stats

get_session_summary

📝 Session summary with git activity

get_session_summary({ sessionID: "editor2-42.2" })
// Returns: git commits, file changes, decisions, anti-pattern warnings

get_lifecycle_status

📈 Decision lifecycle tracking

get_lifecycle_status({ includeCompleted: false })

// Specific decision
get_lifecycle_status({ decisionId: "dec_abc123" })

get_pending_outcomes

⏳ Decisions awaiting feedback

get_pending_outcomes()
// Returns: decisions that need outcome reporting

run_outcome_validation

🔍 Manual outcome validation check

run_outcome_validation()
// Checks for errors, reverts, and stability across all pending decisions

learn_from_outcomes

🧠 Update anti-pattern detection from validated outcomes

learn_from_outcomes()
// Analyzes failed decisions to improve future warnings

init_git_tracking

🔀 Initialize git tracking for current session

init_git_tracking()
// Starts tracking commits, file changes, and line statistics

get_platform_info

🖥️ Platform detection and analytics

get_platform_info({ includeHistory: true })
// Returns: detected platform (cursor, claude-code, etc.),
// environment info, cross-platform learning status

get_ai_generation_rules

📚 AI generation documentation and rules

// Get rules document
get_ai_generation_rules({ document: "rules" })

// Get quick reference
get_ai_generation_rules({ document: "quick-reference", summaryOnly: true })

// Get specific section
get_ai_generation_rules({ document: "rules", section: "Anti-Patterns" })

🔄 Common Workflows

Starting a New Task

// 1. Initialize session
mcp_init({ task: "implement feature X" })

// 2. Get context
preflight({ task: "implement feature X" })

// 3. Check for similar past work
find_similar({ query: "feature X patterns" })

// 4. Check impact
graph_find_impact({ moduleName: "affected-module" })

After Implementation

// 1. Log the decision
log_decision({
  type: "implementation",
  question: "How to implement feature X?",
  chosen: "Used pattern Y",
  reasoning: "Better performance and maintainability",
  confidence: 0.85
})

// 2. Report outcome after testing
report_outcome({
  decisionId: "dec_abc123",
  wasSuccessful: true,
  rating: 4
})

Session Handoff (Multi-Agent)

// When switching agents or resuming work
workflow({ action: "recover_session" })

// For new session orientation
workflow({ action: "discover_project" })

Code Generation for Space Builders

// Generate a complete form
space_generate({
  action: "widget",
  type: "form",
  name: "ContactForm",
  features: ["validation", "submit"]
})

// Generate a data table
space_generate({
  action: "widget",
  type: "table",
  name: "OrdersTable",
  features: ["sorting", "filtering", "pagination", "export"]
})

📅 Version History

Last Updated: January 14, 2026 (Session 42.2) Total Tools: 37 MCP tools documented