MCP Standards

""" Context Optimization System - Practical Examples Demonstrates the intelligent CLAUDE.md optimization system with: 1. Automatic event-driven updates 2. Token reduction (23K → 5K) 3. Learning from manual edits 4. Dynamic context loading """ import asyncio from pathlib import Path import sys # Add src to path sys.path.insert(0, str(Path(__file__).parent.parent / 'src')) from intelligence.context import ContextManager, setup_context_manager from intelligence.memory import PersistentMemory async def demo_basic_setup(): """Demo 1: Basic setup and automatic optimization.""" print("=" * 60) print("Demo 1: Basic Setup and Automatic Optimization") print("=" * 60) # Initialize memory system memory = PersistentMemory( db_path=".claude/demo_memory.db", embedding_model="all-MiniLM-L6-v2" ) # Create context manager manager = ContextManager( project_path=Path.cwd(), memory_system=memory ) # Start watching (don't await - runs in background) await manager.start() print("\n✅ Context manager started") print(f"📁 Watching: {manager.project_path}") print(f"👀 Watching {len(manager.watcher._file_hashes)} files") # Get project analysis print("\n" + "-" * 60) print("Project Analysis:") print("-" * 60) analysis = await manager.analyze_project() print(f"Project Type: {analysis['template_match']['project_type']}") print(f"Template Confidence: {analysis['template_match']['confidence']:.0%}") print(f"Detected Markers: {', '.join(analysis['template_match']['detected_patterns'])}") if analysis['current_metrics']: print(f"\nCurrent CLAUDE.md:") print(f" Token Count: {analysis['current_metrics']['token_count']}") print(f" Section Count: {analysis['current_metrics']['section_count']}") print(f" Optimization Potential: {analysis['current_metrics']['optimization_potential']} tokens") print(f"\nLearned Patterns: {analysis['learned_patterns']}") print(f"High-Confidence Patterns: {analysis['high_confidence_patterns']}") print(f"Available Contexts: {analysis['available_contexts']}") if analysis['recommendations']: print("\n📋 Recommendations:") for rec in analysis['recommendations']: print(f" [{rec['priority'].upper()}] {rec['message']}") print(f" → {rec['action']}") # Clean up await manager.stop() print("\n✅ Manager stopped") async def demo_manual_optimization(): """Demo 2: Manual optimization with metrics.""" print("\n" + "=" * 60) print("Demo 2: Manual Optimization") print("=" * 60) # Quick setup manager = await setup_context_manager( project_path=str(Path.cwd()), auto_start=False # Don't start watcher ) # Check if CLAUDE.md exists claudemd_path = Path.cwd() / 'CLAUDE.md' if not claudemd_path.exists(): print("\n❌ No CLAUDE.md found") print("Creating example CLAUDE.md...") # Create example file example_content = """# Claude Code Configuration ## Core Principles - Evidence > Assumptions - Code > Documentation - Efficiency > Verbosity ## Tool Preferences - Use pytest for testing - Use ruff for linting - Use mypy for type checking ## Workflow Patterns ### Test-Driven Development 1. Write test first 2. Implement feature 3. Refactor ### Code Review - Review before merge - Check tests pass - Verify documentation ## Quality Standards - Code coverage >80% - All tests must pass - No linting errors - Type hints required ## Example Code Patterns ```python # Example: Proper error handling def process_data(data: Dict[str, Any]) -> Result: try: validated = validate(data) return Result.success(validated) except ValidationError as e: return Result.error(str(e)) ``` ## Extended Documentation This is a long section that could be moved to /prime contexts... [Many more paragraphs of detailed documentation] """ claudemd_path.write_text(example_content * 3) # Make it large print(f"✅ Created example CLAUDE.md ({len(example_content * 3)} chars)") # Get initial metrics initial_content = claudemd_path.read_text() initial_tokens = manager.optimizer.estimate_tokens(initial_content) print(f"\n📊 Initial Metrics:") print(f" Token Count: {initial_tokens}") print(f" Target: {manager.optimizer.TOKEN_BUDGET['project']}") print(f" Reduction Needed: {max(0, initial_tokens - manager.optimizer.TOKEN_BUDGET['project'])}") # Optimize print("\n⚙️ Optimizing...") metrics = await manager.optimize_claudemd() print(f"\n✅ Optimization Complete!") print(f" Final Token Count: {metrics.token_count}") print(f" Compression Ratio: {metrics.compression_ratio:.2f}x") print(f" Sections: {metrics.section_count}") print(f" Semantic Density: {metrics.semantic_density:.2f}") # Calculate savings savings = initial_tokens - metrics.token_count savings_percent = (savings / initial_tokens * 100) if initial_tokens > 0 else 0 print(f"\n💰 Token Savings:") print(f" Absolute: {savings} tokens") print(f" Percentage: {savings_percent:.1f}%") async def demo_prime_contexts(): """Demo 3: Dynamic context loading.""" print("\n" + "=" * 60) print("Demo 3: Dynamic Context Loading (/prime commands)") print("=" * 60) manager = await setup_context_manager( project_path=str(Path.cwd()), auto_start=False ) # Show available contexts print("\n📚 Available Prime Contexts:") contexts = manager.list_available_contexts() for ctx in contexts: print(f"\n {ctx['command']}") print(f" Name: {ctx['name']}") print(f" Description: {ctx['description']}") print(f" Tokens: {ctx['tokens']}") if ctx['dependencies']: print(f" Dependencies: {', '.join(ctx['dependencies'])}") # Load a context print("\n" + "-" * 60) print("Loading /prime-bug context...") print("-" * 60) bug_context = await manager.load_prime_context('bug') if bug_context: print(f"\n✅ Loaded bug context ({len(bug_context)} chars)") print("\n--- Context Preview ---") print(bug_context[:500] + "...") # Suggest contexts based on query print("\n" + "-" * 60) print("Context Suggestions for 'performance problem':") print("-" * 60) suggestions = await manager.suggest_contexts("performance problem") for ctx in suggestions: print(f"\n 📌 {ctx.display_name}") print(f" Command: /prime-{ctx.context_id}") print(f" {ctx.description}") print(f" {ctx.token_estimate} tokens") async def demo_learning_system(): """Demo 4: Learning from manual edits.""" print("\n" + "=" * 60) print("Demo 4: Learning from Manual Edits") print("=" * 60) manager = await setup_context_manager( project_path=str(Path.cwd()), auto_start=False ) # Simulate manual edits print("\n📝 Simulating manual edit to CLAUDE.md...") previous_content = """# Claude Code Configuration ## Tool Preferences - Use pip for package management - Use unittest for testing """ current_content = """# Claude Code Configuration ## Tool Preferences - **Use uv for package management** (not pip) - **Use pytest for testing** (not unittest) - Always use ruff for linting """ print("\nPrevious content:") print(previous_content) print("\nNew content:") print(current_content) # Analyze the diff print("\n⚙️ Analyzing diff...") analysis = await manager.learner.analyze_diff( previous_content=previous_content, current_content=current_content, project_path=str(Path.cwd()) ) print(f"\n✅ Analysis Complete!") print(f" Additions: {analysis.additions_count}") print(f" Deletions: {analysis.deletions_count}") print(f" Patterns Detected: {len(analysis.patterns_detected)}") print(f" Token Impact: {analysis.token_impact:+d}") if analysis.patterns_detected: print("\n📋 Learned Patterns:") for pattern in analysis.patterns_detected: print(f"\n Type: {pattern.pattern_type}") print(f" Content: {pattern.content[:100]}") print(f" Confidence: {pattern.confidence:.0%}") print(f" Contexts: {', '.join(pattern.contexts)}") if analysis.preferences_changed: print("\n🔧 Preference Changes:") for pref in analysis.preferences_changed: print(f" - {pref['content']}") if 'preferred' in pref: print(f" Preferred: {pref['preferred']}") if 'avoided' in pref: print(f" Avoided: {pref['avoided']}") # Show how patterns would be auto-applied print("\n" + "-" * 60) print("Auto-Application Demo (after 2+ occurrences):") print("-" * 60) # Simulate second occurrence await manager.learner.analyze_diff( previous_content=current_content, current_content=current_content, # Reinforcement project_path=str(Path.cwd()) ) # Get learned patterns patterns = await manager.learner.get_learned_patterns( min_confidence=0.7 ) print(f"\nLearned Patterns Database: {len(patterns)} patterns") for pattern in patterns[:5]: # Show first 5 print(f"\n Pattern: {pattern.content[:80]}") print(f" Type: {pattern.pattern_type}") print(f" Frequency: {pattern.frequency}") print(f" Confidence: {pattern.confidence:.0%}") print(f" Auto-Apply: {'✅' if pattern.confidence >= 0.8 and pattern.frequency >= 2 else '❌'}") async def demo_suggestions(): """Demo 5: Getting improvement suggestions.""" print("\n" + "=" * 60) print("Demo 5: Improvement Suggestions") print("=" * 60) manager = await setup_context_manager( project_path=str(Path.cwd()), auto_start=False ) # Get suggestions suggestions = await manager.suggest_improvements() if not suggestions: print("\n✅ No suggestions - CLAUDE.md is optimized!") else: print(f"\n📋 Found {len(suggestions)} suggestions:") for i, suggestion in enumerate(suggestions, 1): print(f"\n{i}. [{suggestion['priority'].upper()}] {suggestion['type']}") print(f" {suggestion['suggestion']}") print(f" Confidence: {suggestion['confidence']:.0%}") async def demo_statistics(): """Demo 6: System statistics and monitoring.""" print("\n" + "=" * 60) print("Demo 6: System Statistics") print("=" * 60) manager = await setup_context_manager( project_path=str(Path.cwd()), auto_start=True ) # Let it run for a moment await asyncio.sleep(3) # Get statistics stats = manager.get_statistics() print("\n📊 Context Manager Statistics:") print(f" Running: {stats['running']}") print(f" Optimizations: {stats['manager']['optimizations']}") print(f" Patterns Learned: {stats['manager']['patterns_learned']}") print(f" Contexts Loaded: {stats['manager']['contexts_loaded']}") print(f" Auto Applications: {stats['manager']['auto_applications']}") if 'uptime_seconds' in stats: print(f" Uptime: {stats['uptime_seconds']:.1f}s") if 'watcher' in stats: print("\n📁 File Watcher:") print(f" Events Detected: {stats['watcher']['events_detected']}") print(f" Events Processed: {stats['watcher']['events_processed']}") print(f" Files Tracked: {stats['watcher']['files_tracked']}") print(f" Pending Events: {stats['watcher']['pending_events']}") print(f" Errors: {stats['watcher']['errors']}") print("\n🧠 Learning System:") print(f" Diffs Analyzed: {stats['learner']['diffs_analyzed']}") print(f" Patterns in Database: {stats['learner']['patterns_in_database']}") print(f" High-Confidence Patterns: {stats['learner']['high_confidence_patterns']}") print("\n📚 Prime Context Loader:") print(f" Contexts Loaded: {stats['prime_loader']['contexts_loaded']}") print(f" Cache Hits: {stats['prime_loader']['cache_hits']}") print(f" Cache Misses: {stats['prime_loader']['cache_misses']}") await manager.stop() async def main(): """Run all demos.""" print("\n" + "🚀 " * 30) print("Context Optimization System - Demo Suite") print("🚀 " * 30) try: # Run demos await demo_basic_setup() await demo_manual_optimization() await demo_prime_contexts() await demo_learning_system() await demo_suggestions() await demo_statistics() print("\n" + "=" * 60) print("✅ All demos completed successfully!") print("=" * 60) except Exception as e: print(f"\n❌ Error during demo: {e}") import traceback traceback.print_exc() if __name__ == "__main__": asyncio.run(main())