claude-mpm 3.9.11__py3-none-any.whl → 4.0.4__py3-none-any.whl

claude_mpm/services/agents/loading/agent_profile_loader.py

@@ -19,43 +19,44 @@ This is a consolidated version combining all functionality from the previous
 multi-file implementation for better maintainability.
 """
 
-import os
-import json
-import logging
 import asyncio
 import hashlib
+import json
+import logging
+import os
 from dataclasses import dataclass, field
 from datetime import datetime
 from enum import Enum
 from pathlib import Path
-from typing import Dict, List, Optional, Any, Tuple, Union, Set
+from typing import Any, Dict, List, Optional, Set, Tuple, Union
 
 import yaml
 
 from claude_mpm.core.base_service import BaseService
 from claude_mpm.core.config import Config
-from claude_mpm.core.config_paths import ConfigPaths
-from claude_mpm.services.memory.cache.shared_prompt_cache import SharedPromptCache
+from claude_mpm.core.unified_paths import get_path_manager
 from claude_mpm.services.agents.registry import AgentRegistry
+from claude_mpm.services.memory.cache.shared_prompt_cache import SharedPromptCache
 from claude_mpm.utils.path_operations import path_ops
 
-
 logger = logging.getLogger(__name__)
 
-
 # ============================================================================
 # Data Models
 # ============================================================================
 
+
 class ProfileTier(Enum):
     """Agent profile hierarchy tiers with precedence order."""
+
     PROJECT = "project"  # Highest precedence - project-specific agents
-    USER = "user"       # Medium precedence - user-level agents
-    SYSTEM = "system"   # Lowest precedence - framework/built-in agents
+    USER = "user"  # Medium precedence - user-level agents
+    SYSTEM = "system"  # Lowest precedence - framework/built-in agents
 
 
 class ProfileStatus(Enum):
     """Profile loading status."""
+
     LOADED = "loaded"
     ERROR = "error"
     NOT_FOUND = "not_found"
@@ -65,6 +66,7 @@ class ProfileStatus(Enum):
 @dataclass
 class ImprovedPrompt:
     """Improved prompt data structure."""
+
     prompt_id: str
     agent_name: str
     version: str
@@ -78,6 +80,7 @@ class ImprovedPrompt:
 @dataclass
 class AgentProfile:
     """Complete agent profile with hierarchy information."""
+
     name: str
     role: str
     description: str
@@ -98,57 +101,60 @@ class AgentProfile:
 # Main Service Class
 # ============================================================================
 
+
 class AgentProfileLoader(BaseService):
     """
     Comprehensive agent profile loading service with enhanced prompt integration.
-    
+
     This consolidated version combines all functionality from the previous
     multi-file implementation into a single, maintainable module.
     """
-    
+
     def __init__(self, config: Optional[Config] = None):
         """Initialize the agent profile loader."""
         super().__init__(name="agent_profile_loader", config=config)
-        
+
         # Core configuration
         self.working_directory = Path(os.getcwd())
         self.framework_path = self._detect_framework_path()
         self.user_home = Path.home()
-        
+
         # Tier paths configuration
         self.tier_paths = {
-            ProfileTier.PROJECT: self.working_directory / 'agents',
-            ProfileTier.USER: ConfigPaths.get_user_agents_dir(),
-            ProfileTier.SYSTEM: Path(__file__).parent.parent / 'agents' / 'templates'
+            ProfileTier.PROJECT: self.working_directory / "agents",
+            ProfileTier.USER: get_path_manager().get_user_agents_dir(),
+            ProfileTier.SYSTEM: Path(__file__).parent.parent / "agents" / "templates",
         }
-        
+
         # Remove None values
         self.tier_paths = {k: v for k, v in self.tier_paths.items() if v is not None}
-        
+
         # Profile cache
         self.profile_cache: Dict[str, AgentProfile] = {}
         self.cache_ttl = 3600  # 1 hour
-        
+
         # Service integrations
         self.shared_cache: Optional[SharedPromptCache] = None
         self.agent_registry: Optional[AgentRegistry] = None
-        
+
         # Improved prompts storage
-        self.improved_prompts_path = ConfigPaths.get_user_config_dir() / 'improved_prompts'
+        self.improved_prompts_path = (
+            get_path_manager().get_user_config_dir() / "improved_prompts"
+        )
         self.improved_prompts_path.mkdir(parents=True, exist_ok=True)
-        
+
         # Performance tracking
         self.load_metrics: Dict[str, float] = {}
-        
+
         logger.info(f"AgentProfileLoader initialized successfully")
         logger.info(f"  Working directory: {self.working_directory}")
         logger.info(f"  Framework path: {self.framework_path}")
         logger.info(f"  Tier paths: {list(self.tier_paths.keys())}")
-    
+
     async def _initialize(self) -> None:
         """Initialize the service and its integrations."""
         logger.info("Initializing AgentProfileLoader service...")
-        
+
         # Initialize service integrations
         try:
             self.shared_cache = SharedPromptCache.get_instance()
@@ -156,114 +162,122 @@ class AgentProfileLoader(BaseService):
             logger.info("Successfully initialized service integrations")
         except Exception as e:
             logger.warning(f"Failed to initialize some integrations: {e}")
-        
+
         # Discover and cache initial profiles
         await self._discover_all_profiles()
-        
+
         logger.info("AgentProfileLoader service initialized successfully")
-    
+
     async def _cleanup(self) -> None:
         """Cleanup service resources."""
         logger.info("Cleaning up AgentProfileLoader service...")
-        
+
         # Clear caches
         self.profile_cache.clear()
-        
+
         # Save any pending improved prompts
         await self._save_improved_prompts()
-        
+
         logger.info("AgentProfileLoader service cleaned up")
-    
+
     async def _health_check(self) -> Dict[str, bool]:
         """Perform service health checks."""
         checks = {}
-        
+
         try:
             # Check tier paths
             checks["tier_paths_accessible"] = all(
                 path.exists() for path in self.tier_paths.values()
             )
-            
+
             # Check service integrations
             checks["shared_cache_integration"] = self.shared_cache is not None
             checks["agent_registry_integration"] = self.agent_registry is not None
-            
+
             # Check profile discovery
             profile_count = len(self.profile_cache)
             checks["profiles_loaded"] = profile_count > 0
-            
+
             # Check improved prompts storage
             checks["improved_prompts_storage"] = self.improved_prompts_path.exists()
-            
+
         except Exception as e:
             logger.error(f"Health check failed: {e}")
             checks["health_check_error"] = False
-        
+
         return checks
-    
+
     # ========================================================================
     # Core Profile Loading
     # ========================================================================
-    
-    async def load_agent_profile(self, agent_name: str, 
-                               use_cache: bool = True) -> Optional[AgentProfile]:
+
+    async def load_agent_profile(
+        self, agent_name: str, use_cache: bool = True
+    ) -> Optional[AgentProfile]:
         """
         Load agent profile with three-tier hierarchy precedence.
-        
+
         Args:
             agent_name: Name of the agent to load
             use_cache: Whether to use cached profile if available
-            
+
         Returns:
             AgentProfile or None if not found
         """
         start_time = asyncio.get_event_loop().time()
-        
+
         # Check cache first
         if use_cache and agent_name in self.profile_cache:
             profile = self.profile_cache[agent_name]
             if (datetime.now() - profile.loaded_at).seconds < self.cache_ttl:
-                self.load_metrics[f"{agent_name}_cache_hit"] = asyncio.get_event_loop().time() - start_time
+                self.load_metrics[f"{agent_name}_cache_hit"] = (
+                    asyncio.get_event_loop().time() - start_time
+                )
                 return profile
-        
+
         # Load profile with tier precedence
         profile = await self._load_profile_with_precedence(agent_name)
-        
+
         if profile:
             # Load improved prompts
             profile.improved_prompts = await self._load_improved_prompts(agent_name)
-            
+
             # Cache the profile
             self.profile_cache[agent_name] = profile
-            
+
             # Update metrics
-            self.load_metrics[f"{agent_name}_load_time"] = asyncio.get_event_loop().time() - start_time
-            
+            self.load_metrics[f"{agent_name}_load_time"] = (
+                asyncio.get_event_loop().time() - start_time
+            )
+
             # Integrate with SharedPromptCache if available
             if self.shared_cache and profile.status == ProfileStatus.LOADED:
                 cache_key = f"agent_profile_{agent_name}"
                 await self.shared_cache.set_cached_result(cache_key, profile)
                 profile.cache_key = cache_key
-        
+
         return profile
-    
-    async def _load_profile_with_precedence(self, agent_name: str) -> Optional[AgentProfile]:
+
+    async def _load_profile_with_precedence(
+        self, agent_name: str
+    ) -> Optional[AgentProfile]:
         """Load profile following tier precedence: Project → User → System."""
         for tier in [ProfileTier.PROJECT, ProfileTier.USER, ProfileTier.SYSTEM]:
             if tier not in self.tier_paths:
                 continue
-                
+
             profile = await self._load_profile_from_tier(agent_name, tier)
             if profile and profile.status == ProfileStatus.LOADED:
                 return profile
-        
+
         return None
-    
-    async def _load_profile_from_tier(self, agent_name: str, 
-                                    tier: ProfileTier) -> Optional[AgentProfile]:
+
+    async def _load_profile_from_tier(
+        self, agent_name: str, tier: ProfileTier
+    ) -> Optional[AgentProfile]:
         """Load profile from specific tier."""
         tier_path = self.tier_paths[tier]
-        
+
         # Try different file formats and naming conventions
         # Check .md files first (Claude Code format), then fall back to YAML/JSON
         possible_files = [
@@ -280,380 +294,396 @@ class AgentProfileLoader(BaseService):
             tier_path / f"{agent_name}-agent.yml",
             tier_path / f"{agent_name}-agent.json",
         ]
-        
+
         for file_path in possible_files:
             if file_path.exists():
                 return await self._parse_profile_file(file_path, tier)
-        
+
         return None
-    
-    async def _parse_profile_file(self, file_path: Path, 
-                                tier: ProfileTier) -> Optional[AgentProfile]:
+
+    async def _parse_profile_file(
+        self, file_path: Path, tier: ProfileTier
+    ) -> Optional[AgentProfile]:
         """Parse agent profile from file."""
         try:
             # Read file content
             content = file_path.read_text()
-            
+
             # Parse based on file extension
-            if file_path.suffix == '.md':
+            if file_path.suffix == ".md":
                 # Parse markdown with YAML frontmatter
                 data, instructions = self._parse_markdown_with_frontmatter(content)
-            elif file_path.suffix in ['.yaml', '.yml']:
+            elif file_path.suffix in [".yaml", ".yml"]:
                 data = yaml.safe_load(content)
-                instructions = data.get('instructions', '')
-            elif file_path.suffix == '.json':
+                instructions = data.get("instructions", "")
+            elif file_path.suffix == ".json":
                 data = json.loads(content)
-                instructions = data.get('instructions', '')
+                instructions = data.get("instructions", "")
             else:
                 # Try to parse as YAML first, then JSON
                 try:
                     data = yaml.safe_load(content)
-                    instructions = data.get('instructions', '')
+                    instructions = data.get("instructions", "")
                 except:
                     data = json.loads(content)
-                    instructions = data.get('instructions', '')
-            
+                    instructions = data.get("instructions", "")
+
             # Create profile
             profile = AgentProfile(
-                name=data.get('name', file_path.stem),
-                role=data.get('role', 'agent'),
-                description=data.get('description', ''),
+                name=data.get("name", file_path.stem),
+                role=data.get("role", "agent"),
+                description=data.get("description", ""),
                 tier=tier,
                 source_path=str(file_path),
                 instructions=instructions,
-                capabilities=data.get('capabilities', []),
-                constraints=data.get('constraints', []),
-                metadata=data.get('metadata', {}),
-                status=ProfileStatus.LOADED
+                capabilities=data.get("capabilities", []),
+                constraints=data.get("constraints", []),
+                metadata=data.get("metadata", {}),
+                status=ProfileStatus.LOADED,
             )
-            
+
             return profile
-            
+
         except Exception as e:
             logger.error(f"Error parsing profile {file_path}: {e}")
             return AgentProfile(
                 name=file_path.stem,
-                role='error',
-                description='Failed to load profile',
+                role="error",
+                description="Failed to load profile",
                 tier=tier,
                 source_path=str(file_path),
                 status=ProfileStatus.ERROR,
-                error=str(e)
+                error=str(e),
             )
-    
-    def _parse_markdown_with_frontmatter(self, content: str) -> Tuple[Dict[str, Any], str]:
+
+    def _parse_markdown_with_frontmatter(
+        self, content: str
+    ) -> Tuple[Dict[str, Any], str]:
         """
         Parse markdown file with YAML frontmatter.
-        
+
         Args:
             content: Markdown content with YAML frontmatter
-            
+
         Returns:
             Tuple of (frontmatter_data, markdown_content)
         """
         import re
-        
+
         # Check if content starts with YAML frontmatter
-        if not content.strip().startswith('---'):
+        if not content.strip().startswith("---"):
             # No frontmatter, treat entire content as instructions
-            return {'name': 'unknown', 'description': 'No frontmatter found'}, content
-        
+            return {"name": "unknown", "description": "No frontmatter found"}, content
+
         # Split frontmatter and content
-        parts = re.split(r'^---\s*$', content, 2, re.MULTILINE)
-        
+        parts = re.split(r"^---\s*$", content, 2, re.MULTILINE)
+
         if len(parts) < 3:
             # Invalid frontmatter structure
-            return {'name': 'unknown', 'description': 'Invalid frontmatter'}, content
-        
+            return {"name": "unknown", "description": "Invalid frontmatter"}, content
+
         # Parse YAML frontmatter
         frontmatter_text = parts[1].strip()
         markdown_content = parts[2].strip()
-        
+
         try:
             frontmatter_data = yaml.safe_load(frontmatter_text)
             if not isinstance(frontmatter_data, dict):
-                frontmatter_data = {'name': 'unknown', 'description': 'Invalid frontmatter format'}
+                frontmatter_data = {
+                    "name": "unknown",
+                    "description": "Invalid frontmatter format",
+                }
         except Exception as e:
             logger.error(f"Error parsing YAML frontmatter: {e}")
-            frontmatter_data = {'name': 'unknown', 'description': f'YAML parse error: {e}'}
-        
+            frontmatter_data = {
+                "name": "unknown",
+                "description": f"YAML parse error: {e}",
+            }
+
         return frontmatter_data, markdown_content
-    
+
     # ========================================================================
     # Profile Discovery
     # ========================================================================
-    
+
     async def _discover_all_profiles(self) -> Dict[ProfileTier, List[str]]:
         """Discover all available agent profiles across tiers."""
         discovered = {}
-        
+
         for tier, tier_path in self.tier_paths.items():
             if not tier_path.exists():
                 continue
-                
+
             agents = []
             # Check for .md files (Claude Code format) and YAML/JSON files
-            file_patterns = ['*.md', '*.yaml', '*.yml', '*.json']
+            file_patterns = ["*.md", "*.yaml", "*.yml", "*.json"]
             for pattern in file_patterns:
                 for file_path in tier_path.glob(pattern):
                     agent_name = file_path.stem
                     # Remove common suffixes
-                    if agent_name.endswith('_agent'):
+                    if agent_name.endswith("_agent"):
                         agent_name = agent_name[:-6]
-                    elif agent_name.endswith('-agent'):
+                    elif agent_name.endswith("-agent"):
                         agent_name = agent_name[:-6]
-                    
+
                     if agent_name not in agents:
                         agents.append(agent_name)
-            
+
             discovered[tier] = agents
             logger.debug(f"Discovered {len(agents)} agents in {tier.value} tier")
-        
+
         return discovered
-    
-    async def get_available_agents(self, tier: Optional[ProfileTier] = None) -> List[str]:
+
+    async def get_available_agents(
+        self, tier: Optional[ProfileTier] = None
+    ) -> List[str]:
         """Get list of available agents, optionally filtered by tier."""
         discovered = await self._discover_all_profiles()
-        
+
         if tier:
             return discovered.get(tier, [])
-        
+
         # Combine all tiers, removing duplicates
         all_agents = set()
         for agents in discovered.values():
             all_agents.update(agents)
-        
+
         return sorted(list(all_agents))
-    
+
     # ========================================================================
     # Improved Prompts Management
     # ========================================================================
-    
+
     async def _load_improved_prompts(self, agent_name: str) -> List[ImprovedPrompt]:
         """Load improved prompts for an agent."""
         prompts = []
         prompt_file = self.improved_prompts_path / f"{agent_name}_prompts.json"
-        
+
         if prompt_file.exists():
             try:
-                with open(prompt_file, 'r') as f:
+                with open(prompt_file, "r") as f:
                     data = json.load(f)
                     for prompt_data in data:
                         prompt = ImprovedPrompt(
-                            prompt_id=prompt_data['prompt_id'],
-                            agent_name=prompt_data['agent_name'],
-                            version=prompt_data['version'],
-                            content=prompt_data['content'],
-                            metrics=prompt_data.get('metrics', {}),
-                            metadata=prompt_data.get('metadata', {}),
-                            created_at=datetime.fromisoformat(prompt_data['created_at']),
-                            validated=prompt_data.get('validated', False)
+                            prompt_id=prompt_data["prompt_id"],
+                            agent_name=prompt_data["agent_name"],
+                            version=prompt_data["version"],
+                            content=prompt_data["content"],
+                            metrics=prompt_data.get("metrics", {}),
+                            metadata=prompt_data.get("metadata", {}),
+                            created_at=datetime.fromisoformat(
+                                prompt_data["created_at"]
+                            ),
+                            validated=prompt_data.get("validated", False),
                         )
                         prompts.append(prompt)
             except Exception as e:
                 logger.error(f"Error loading improved prompts for {agent_name}: {e}")
-        
+
         return prompts
-    
-    async def save_improved_prompt(self, agent_name: str, 
-                                 prompt: ImprovedPrompt) -> bool:
+
+    async def save_improved_prompt(
+        self, agent_name: str, prompt: ImprovedPrompt
+    ) -> bool:
         """Save an improved prompt for an agent."""
         try:
             # Load existing prompts
             prompts = await self._load_improved_prompts(agent_name)
-            
+
             # Add or update prompt
             existing_idx = None
             for idx, existing in enumerate(prompts):
                 if existing.prompt_id == prompt.prompt_id:
                     existing_idx = idx
                     break
-            
+
             if existing_idx is not None:
                 prompts[existing_idx] = prompt
             else:
                 prompts.append(prompt)
-            
+
             # Save to file
             prompt_file = self.improved_prompts_path / f"{agent_name}_prompts.json"
-            with open(prompt_file, 'w') as f:
-                json.dump([
-                    {
-                        'prompt_id': p.prompt_id,
-                        'agent_name': p.agent_name,
-                        'version': p.version,
-                        'content': p.content,
-                        'metrics': p.metrics,
-                        'metadata': p.metadata,
-                        'created_at': p.created_at.isoformat(),
-                        'validated': p.validated
-                    }
-                    for p in prompts
-                ], f, indent=2)
-            
+            with open(prompt_file, "w") as f:
+                json.dump(
+                    [
+                        {
+                            "prompt_id": p.prompt_id,
+                            "agent_name": p.agent_name,
+                            "version": p.version,
+                            "content": p.content,
+                            "metrics": p.metrics,
+                            "metadata": p.metadata,
+                            "created_at": p.created_at.isoformat(),
+                            "validated": p.validated,
+                        }
+                        for p in prompts
+                    ],
+                    f,
+                    indent=2,
+                )
+
             return True
-            
+
         except Exception as e:
             logger.error(f"Error saving improved prompt: {e}")
             return False
-    
+
     async def _save_improved_prompts(self) -> None:
         """Save all pending improved prompts."""
         for agent_name, profile in self.profile_cache.items():
             if profile.improved_prompts:
                 for prompt in profile.improved_prompts:
                     await self.save_improved_prompt(agent_name, prompt)
-    
+
     # ========================================================================
     # Task Integration
     # ========================================================================
-    
-    async def enhance_task_creation(self, agent_name: str, 
-                                  task_params: Dict[str, Any]) -> Dict[str, Any]:
+
+    async def enhance_task_creation(
+        self, agent_name: str, task_params: Dict[str, Any]
+    ) -> Dict[str, Any]:
         """
         Enhance Task Tool subprocess creation with agent profile data.
-        
+
         Args:
             agent_name: Name of the agent
             task_params: Original task parameters
-            
+
         Returns:
             Enhanced task parameters
         """
         profile = await self.load_agent_profile(agent_name)
-        
+
         if not profile:
             logger.warning(f"No profile found for agent {agent_name}")
             return task_params
-        
+
         # Enhance with profile data
         enhanced = task_params.copy()
-        
+
         # Add agent metadata
-        enhanced['agent_metadata'] = {
-            'name': profile.name,
-            'role': profile.role,
-            'tier': profile.tier.value,
-            'capabilities': profile.capabilities,
-            'constraints': profile.constraints
+        enhanced["agent_metadata"] = {
+            "name": profile.name,
+            "role": profile.role,
+            "tier": profile.tier.value,
+            "capabilities": profile.capabilities,
+            "constraints": profile.constraints,
         }
-        
+
         # Add improved prompts if available
         if profile.improved_prompts:
             best_prompt = max(
-                profile.improved_prompts,
-                key=lambda p: p.metrics.get('success_rate', 0)
+                profile.improved_prompts, key=lambda p: p.metrics.get("success_rate", 0)
             )
-            enhanced['improved_prompt'] = {
-                'content': best_prompt.content,
-                'version': best_prompt.version,
-                'metrics': best_prompt.metrics
+            enhanced["improved_prompt"] = {
+                "content": best_prompt.content,
+                "version": best_prompt.version,
+                "metrics": best_prompt.metrics,
             }
-        
+
         # Add custom instructions if available
         if profile.instructions:
-            enhanced['additional_instructions'] = profile.instructions
-        
+            enhanced["additional_instructions"] = profile.instructions
+
         return enhanced
-    
+
     # ========================================================================
     # Metrics and Validation
     # ========================================================================
-    
+
     async def validate_profile(self, agent_name: str) -> Dict[str, Any]:
         """Validate agent profile structure and content."""
         profile = await self.load_agent_profile(agent_name)
-        
+
         if not profile:
-            return {'valid': False, 'error': 'Profile not found'}
-        
-        validation_results = {
-            'valid': True,
-            'warnings': [],
-            'errors': []
-        }
-        
+            return {"valid": False, "error": "Profile not found"}
+
+        validation_results = {"valid": True, "warnings": [], "errors": []}
+
         # Check required fields
         if not profile.name:
-            validation_results['errors'].append('Missing agent name')
-            validation_results['valid'] = False
-        
+            validation_results["errors"].append("Missing agent name")
+            validation_results["valid"] = False
+
         if not profile.role:
-            validation_results['errors'].append('Missing agent role')
-            validation_results['valid'] = False
-        
+            validation_results["errors"].append("Missing agent role")
+            validation_results["valid"] = False
+
         # Check profile completeness
         if not profile.description:
-            validation_results['warnings'].append('Missing agent description')
-        
+            validation_results["warnings"].append("Missing agent description")
+
         if not profile.capabilities:
-            validation_results['warnings'].append('No capabilities defined')
-        
+            validation_results["warnings"].append("No capabilities defined")
+
         if not profile.constraints:
-            validation_results['warnings'].append('No constraints defined')
-        
+            validation_results["warnings"].append("No constraints defined")
+
         # Validate improved prompts
         for prompt in profile.improved_prompts:
             if not prompt.validated:
-                validation_results['warnings'].append(
-                    f'Unvalidated prompt: {prompt.prompt_id}'
+                validation_results["warnings"].append(
+                    f"Unvalidated prompt: {prompt.prompt_id}"
                 )
-        
+
         return validation_results
-    
+
     async def get_profile_metrics(self) -> Dict[str, Any]:
         """Get comprehensive profile loading metrics."""
         metrics = {
-            'cache_size': len(self.profile_cache),
-            'load_metrics': self.load_metrics,
-            'tier_stats': {}
+            "cache_size": len(self.profile_cache),
+            "load_metrics": self.load_metrics,
+            "tier_stats": {},
         }
-        
+
         # Count profiles by tier
         for profile in self.profile_cache.values():
             tier = profile.tier.value
-            metrics['tier_stats'][tier] = metrics['tier_stats'].get(tier, 0) + 1
-        
+            metrics["tier_stats"][tier] = metrics["tier_stats"].get(tier, 0) + 1
+
         # Calculate average load times
         if self.load_metrics:
-            load_times = [v for k, v in self.load_metrics.items() if k.endswith('_load_time')]
+            load_times = [
+                v for k, v in self.load_metrics.items() if k.endswith("_load_time")
+            ]
             if load_times:
-                metrics['avg_load_time'] = sum(load_times) / len(load_times)
-        
+                metrics["avg_load_time"] = sum(load_times) / len(load_times)
+
         # Count improved prompts
         total_prompts = sum(
             len(p.improved_prompts) for p in self.profile_cache.values()
         )
-        metrics['total_improved_prompts'] = total_prompts
-        
+        metrics["total_improved_prompts"] = total_prompts
+
         return metrics
-    
+
     # ========================================================================
     # Utility Methods
     # ========================================================================
-    
+
     def _detect_framework_path(self) -> Optional[Path]:
         """Detect the framework path for system-level agents."""
         possible_paths = [
-            self.working_directory / 'framework',
-            self.working_directory / 'src' / 'claude_mpm' / 'framework',
-            Path(__file__).parent.parent / 'framework'
+            self.working_directory / "framework",
+            self.working_directory / "src" / "claude_mpm" / "framework",
+            Path(__file__).parent.parent / "framework",
         ]
-        
+
         for path in possible_paths:
-            if path.exists() and (path / 'agent-roles').exists():
+            if path.exists() and (path / "agent-roles").exists():
                 return path
-        
+
         return None
-    
+
     def invalidate_cache(self, agent_name: Optional[str] = None) -> None:
         """Invalidate profile cache."""
         if agent_name:
             self.profile_cache.pop(agent_name, None)
         else:
             self.profile_cache.clear()
-    
+
     async def reload_profile(self, agent_name: str) -> Optional[AgentProfile]:
         """Force reload of agent profile, bypassing cache."""
         self.invalidate_cache(agent_name)
-        return await self.load_agent_profile(agent_name, use_cache=False)
+        return await self.load_agent_profile(agent_name, use_cache=False)