claude-mpm 4.24.0__py3-none-any.whl → 5.0.9__py3-none-any.whl

claude_mpm/core/framework/__init__.py

@@ -4,22 +4,9 @@ This module provides the modular framework loading system with specialized compo
 for handling different aspects of framework initialization and management.
 """
 
-from .formatters import (
-    CapabilityGenerator,
-    ContentFormatter,
-    ContextGenerator,
-)
-from .loaders import (
-    AgentLoader,
-    FileLoader,
-    InstructionLoader,
-    PackagedLoader,
-)
-from .processors import (
-    MemoryProcessor,
-    MetadataProcessor,
-    TemplateProcessor,
-)
+from .formatters import CapabilityGenerator, ContentFormatter, ContextGenerator
+from .loaders import AgentLoader, FileLoader, InstructionLoader, PackagedLoader
+from .processors import MemoryProcessor, MetadataProcessor, TemplateProcessor
 
 __all__ = [
     "AgentLoader",

claude_mpm/core/framework/loaders/file_loader.py

@@ -66,6 +66,53 @@ class FileLoader:
             if "INSTRUCTIONS.md" in str(file_path):
                 self.framework_last_modified = timestamp
 
+    def _load_tier_file(
+        self,
+        filename: str,
+        current_dir: Path,
+        framework_path: Optional[Path] = None,
+        include_system: bool = False,
+    ) -> tuple[Optional[str], Optional[str]]:
+        """Load file with tier precedence: project → user → system.
+
+        Args:
+            filename: Name of file to load (e.g., "INSTRUCTIONS.md")
+            current_dir: Current working directory for project-level
+            framework_path: Path to framework installation for system-level
+            include_system: Whether to check system-level path
+
+        Returns:
+            Tuple of (content, level) where level is 'project', 'user', 'system', or None
+        """
+        # Check project-level (highest priority)
+        project_path = current_dir / ".claude-mpm" / filename
+        if project_path.exists():
+            loaded_content = self.try_load_file(
+                project_path, f"project-specific {filename}"
+            )
+            if loaded_content:
+                self.logger.info(f"Using project-specific {filename} from .claude-mpm/")
+                return loaded_content, "project"
+
+        # Check user-level (medium priority)
+        user_path = Path.home() / ".claude-mpm" / filename
+        if user_path.exists():
+            loaded_content = self.try_load_file(user_path, f"user-specific {filename}")
+            if loaded_content:
+                self.logger.info(f"Using user-specific {filename} from ~/.claude-mpm/")
+                return loaded_content, "user"
+
+        # Check system-level (lowest priority)
+        if include_system and framework_path and framework_path != Path("__PACKAGED__"):
+            system_path = framework_path / "src" / "claude_mpm" / "agents" / filename
+            if system_path.exists():
+                loaded_content = self.try_load_file(system_path, f"system {filename}")
+                if loaded_content:
+                    self.logger.info(f"Using system {filename}")
+                    return loaded_content, "system"
+
+        return None, None
+
     def load_instructions_file(
         self, current_dir: Path
     ) -> tuple[Optional[str], Optional[str]]:
@@ -82,31 +129,7 @@ class FileLoader:
         Returns:
             Tuple of (content, level) where level is 'project', 'user', or None
         """
-        # Check for project-specific INSTRUCTIONS.md first
-        project_instructions_path = current_dir / ".claude-mpm" / "INSTRUCTIONS.md"
-        if project_instructions_path.exists():
-            loaded_content = self.try_load_file(
-                project_instructions_path, "project-specific INSTRUCTIONS.md"
-            )
-            if loaded_content:
-                self.logger.info(
-                    "Using project-specific PM instructions from .claude-mpm/INSTRUCTIONS.md"
-                )
-                return loaded_content, "project"
-
-        # Check for user-specific INSTRUCTIONS.md
-        user_instructions_path = Path.home() / ".claude-mpm" / "INSTRUCTIONS.md"
-        if user_instructions_path.exists():
-            loaded_content = self.try_load_file(
-                user_instructions_path, "user-specific INSTRUCTIONS.md"
-            )
-            if loaded_content:
-                self.logger.info(
-                    "Using user-specific PM instructions from ~/.claude-mpm/INSTRUCTIONS.md"
-                )
-                return loaded_content, "user"
-
-        return None, None
+        return self._load_tier_file("INSTRUCTIONS.md", current_dir)
 
     def load_workflow_file(
         self, current_dir: Path, framework_path: Path
@@ -126,44 +149,9 @@ class FileLoader:
         Returns:
             Tuple of (content, level) where level is 'project', 'user', 'system', or None
         """
-        # Check for project-specific WORKFLOW.md first (highest priority)
-        project_workflow_path = current_dir / ".claude-mpm" / "WORKFLOW.md"
-        if project_workflow_path.exists():
-            loaded_content = self.try_load_file(
-                project_workflow_path, "project-specific WORKFLOW.md"
-            )
-            if loaded_content:
-                self.logger.info(
-                    "Using project-specific workflow instructions from .claude-mpm/WORKFLOW.md"
-                )
-                return loaded_content, "project"
-
-        # Check for user-specific WORKFLOW.md (medium priority)
-        user_workflow_path = Path.home() / ".claude-mpm" / "WORKFLOW.md"
-        if user_workflow_path.exists():
-            loaded_content = self.try_load_file(
-                user_workflow_path, "user-specific WORKFLOW.md"
-            )
-            if loaded_content:
-                self.logger.info(
-                    "Using user-specific workflow instructions from ~/.claude-mpm/WORKFLOW.md"
-                )
-                return loaded_content, "user"
-
-        # Fall back to system workflow (lowest priority)
-        if framework_path and framework_path != Path("__PACKAGED__"):
-            system_workflow_path = (
-                framework_path / "src" / "claude_mpm" / "agents" / "WORKFLOW.md"
-            )
-            if system_workflow_path.exists():
-                loaded_content = self.try_load_file(
-                    system_workflow_path, "system WORKFLOW.md"
-                )
-                if loaded_content:
-                    self.logger.info("Using system workflow instructions")
-                    return loaded_content, "system"
-
-        return None, None
+        return self._load_tier_file(
+            "WORKFLOW.md", current_dir, framework_path, include_system=True
+        )
 
     def load_memory_file(
         self, current_dir: Path, framework_path: Path
@@ -183,41 +171,6 @@ class FileLoader:
         Returns:
             Tuple of (content, level) where level is 'project', 'user', 'system', or None
         """
-        # Check for project-specific MEMORY.md first (highest priority)
-        project_memory_path = current_dir / ".claude-mpm" / "MEMORY.md"
-        if project_memory_path.exists():
-            loaded_content = self.try_load_file(
-                project_memory_path, "project-specific MEMORY.md"
-            )
-            if loaded_content:
-                self.logger.info(
-                    "Using project-specific memory instructions from .claude-mpm/MEMORY.md"
-                )
-                return loaded_content, "project"
-
-        # Check for user-specific MEMORY.md (medium priority)
-        user_memory_path = Path.home() / ".claude-mpm" / "MEMORY.md"
-        if user_memory_path.exists():
-            loaded_content = self.try_load_file(
-                user_memory_path, "user-specific MEMORY.md"
-            )
-            if loaded_content:
-                self.logger.info(
-                    "Using user-specific memory instructions from ~/.claude-mpm/MEMORY.md"
-                )
-                return loaded_content, "user"
-
-        # Fall back to system memory instructions (lowest priority)
-        if framework_path and framework_path != Path("__PACKAGED__"):
-            system_memory_path = (
-                framework_path / "src" / "claude_mpm" / "agents" / "MEMORY.md"
-            )
-            if system_memory_path.exists():
-                loaded_content = self.try_load_file(
-                    system_memory_path, "system MEMORY.md"
-                )
-                if loaded_content:
-                    self.logger.info("Using system memory instructions")
-                    return loaded_content, "system"
-
-        return None, None
+        return self._load_tier_file(
+            "MEMORY.md", current_dir, framework_path, include_system=True
+        )

claude_mpm/core/framework/loaders/instruction_loader.py

@@ -91,10 +91,28 @@ class InstructionLoader:
     def _load_filesystem_framework_instructions(self, content: Dict[str, Any]) -> None:
         """Load framework instructions from filesystem.
 
+        Priority order:
+        1. Deployed compiled file: .claude-mpm/PM_INSTRUCTIONS_DEPLOYED.md (HIGHEST PRIORITY)
+        2. Source file (development): src/claude_mpm/agents/PM_INSTRUCTIONS.md
+        3. Legacy file (backward compat): src/claude_mpm/agents/INSTRUCTIONS.md
+
         Args:
             content: Dictionary to update with framework instructions
         """
-        # Try new consolidated PM_INSTRUCTIONS.md first, fall back to INSTRUCTIONS.md
+        # PRIORITY 1: Check for compiled/deployed version in .claude-mpm/
+        # This is the merged PM_INSTRUCTIONS.md + WORKFLOW.md + MEMORY.md
+        deployed_path = self.current_dir / ".claude-mpm" / "PM_INSTRUCTIONS_DEPLOYED.md"
+        if deployed_path.exists():
+            loaded_content = self.file_loader.try_load_file(
+                deployed_path, "deployed PM_INSTRUCTIONS_DEPLOYED.md"
+            )
+            if loaded_content:
+                content["framework_instructions"] = loaded_content
+                content["loaded"] = True
+                self.logger.info("Loaded PM_INSTRUCTIONS_DEPLOYED.md from .claude-mpm/")
+                return  # Stop here - deployed version takes precedence
+
+        # PRIORITY 2: Development mode - load from source PM_INSTRUCTIONS.md
         pm_instructions_path = (
             self.framework_path / "src" / "claude_mpm" / "agents" / "PM_INSTRUCTIONS.md"
         )
@@ -102,16 +120,18 @@ class InstructionLoader:
             self.framework_path / "src" / "claude_mpm" / "agents" / "INSTRUCTIONS.md"
         )
 
-        # Try loading new consolidated file first
+        # Try loading new consolidated file
         if pm_instructions_path.exists():
             loaded_content = self.file_loader.try_load_file(
-                pm_instructions_path, "consolidated PM_INSTRUCTIONS.md"
+                pm_instructions_path, "source PM_INSTRUCTIONS.md (development mode)"
             )
             if loaded_content:
                 content["framework_instructions"] = loaded_content
                 content["loaded"] = True
-                self.logger.info("Loaded consolidated PM_INSTRUCTIONS.md")
-        # Fall back to legacy file for backward compatibility
+                self.logger.warning(
+                    "Using source PM_INSTRUCTIONS.md - deployed version not found"
+                )
+        # PRIORITY 3: Fall back to legacy file for backward compatibility
         elif framework_instructions_path.exists():
             loaded_content = self.file_loader.try_load_file(
                 framework_instructions_path, "framework INSTRUCTIONS.md (legacy)"

claude_mpm/core/hook_error_memory.py

@@ -0,0 +1,381 @@
+"""Hook Error Memory System.
+
+This module provides error detection and memory for hook execution to prevent
+repeated errors and provide helpful diagnostics.
+
+WHY this is needed:
+- Hook processor can encounter transient or persistent errors
+- Repeated failing commands waste resources and clutter logs
+- Users need actionable suggestions to fix configuration issues
+- System should learn from errors and prevent repetition
+
+DESIGN DECISION: Store errors in JSON file rather than database because:
+- Simple, human-readable format
+- Easy to inspect and manually clear
+- No additional dependencies
+- Fast read/write for small datasets
+- Users can easily delete to retry failed commands
+"""
+
+import json
+import re
+from datetime import datetime, timezone
+from pathlib import Path
+from typing import Any, Dict, Optional
+
+from ..core.logger import get_logger
+
+
+class HookErrorMemory:
+    """Tracks and prevents repeated hook execution errors.
+
+    WHY this design:
+    - Detects common error patterns automatically
+    - Stores error history to prevent repetition
+    - Provides actionable fix suggestions
+    - Allows manual retry by clearing memory
+    - Minimal performance overhead (<1ms per check)
+    """
+
+    # Error pattern definitions with detection regexes
+    ERROR_PATTERNS = [
+        (r"no such file or directory[:\s]+(.+?)(?:\n|$)", "file_not_found"),
+        (r"command not found[:\s]+(.+?)(?:\n|$)", "command_not_found"),
+        (r"permission denied[:\s]+(.+?)(?:\n|$)", "permission_denied"),
+        (r"syntax error", "syntax_error"),
+        (r"Error:\s*\(eval\):(\d+):\s*(.+?)(?:\n|$)", "eval_error"),
+        (r"Error:\s*(.+?)(?:\n|$)", "general_error"),
+    ]
+
+    def __init__(self, memory_file: Optional[Path] = None):
+        """Initialize hook error memory.
+
+        Args:
+            memory_file: Path to memory file (default: .claude-mpm/hook_errors.json)
+        """
+        self.logger = get_logger("hook_error_memory")
+
+        # Use default location if not specified
+        if memory_file is None:
+            memory_file = Path.cwd() / ".claude-mpm" / "hook_errors.json"
+
+        self.memory_file = memory_file
+        self.errors: Dict[str, Any] = self._load_errors()
+
+    def _load_errors(self) -> Dict[str, Any]:
+        """Load previously encountered errors from disk.
+
+        Returns:
+            Dictionary of error records
+        """
+        if not self.memory_file.exists():
+            return {}
+
+        try:
+            content = self.memory_file.read_text()
+            if not content.strip():
+                return {}
+            return json.loads(content)
+        except json.JSONDecodeError as e:
+            self.logger.warning(f"Failed to parse error memory file: {e}")
+            return {}
+        except Exception as e:
+            self.logger.error(f"Error loading error memory: {e}")
+            return {}
+
+    def _save_errors(self):
+        """Persist errors to disk."""
+        try:
+            # Ensure directory exists
+            self.memory_file.parent.mkdir(parents=True, exist_ok=True)
+
+            # Write with pretty formatting for human readability
+            self.memory_file.write_text(json.dumps(self.errors, indent=2))
+        except Exception as e:
+            self.logger.error(f"Failed to save error memory: {e}")
+
+    def detect_error(
+        self, output: str, stderr: str, returncode: int
+    ) -> Optional[Dict[str, str]]:
+        """Detect if output contains an error.
+
+        WHY check both stdout and stderr:
+        - Some commands write errors to stdout
+        - Some write to stderr
+        - Return code alone isn't enough (some hooks return non-zero on purpose)
+
+        Args:
+            output: Standard output from command
+            stderr: Standard error from command
+            returncode: Exit code from command
+
+        Returns:
+            Dict with error info if detected, None otherwise
+        """
+        # Combine output sources for comprehensive error detection
+        combined_output = f"{output}\n{stderr}"
+
+        # Try each pattern in order of specificity
+        for pattern, error_type in self.ERROR_PATTERNS:
+            match = re.search(pattern, combined_output, re.IGNORECASE | re.MULTILINE)
+            if match:
+                # Extract details from the match
+                details = match.group(1) if match.groups() else match.group(0)
+
+                return {
+                    "type": error_type,
+                    "pattern": pattern,
+                    "match": match.group(0).strip(),
+                    "details": details.strip() if details else "",
+                    "returncode": returncode,
+                }
+
+        # If no pattern matched but returncode is non-zero, record as generic error
+        if returncode != 0 and combined_output.strip():
+            return {
+                "type": "unknown_error",
+                "pattern": "non-zero exit code",
+                "match": f"Exit code: {returncode}",
+                "details": combined_output[:200].strip(),  # First 200 chars
+                "returncode": returncode,
+            }
+
+        return None
+
+    def record_error(self, error_info: Dict[str, str], hook_type: str):
+        """Record an error to prevent future repetition.
+
+        WHY use composite key:
+        - Same error type can occur with different details
+        - Want to track specific error instances
+        - Hook type context helps with diagnosis
+
+        Args:
+            error_info: Error information from detect_error()
+            hook_type: Type of hook that failed (e.g., "PreToolUse")
+        """
+        # Create unique key for this error
+        key = f"{error_info['type']}:{hook_type}:{error_info['details']}"
+
+        now = datetime.now(timezone.utc).isoformat()
+
+        if key in self.errors:
+            # Update existing error
+            self.errors[key]["count"] += 1
+            self.errors[key]["last_seen"] = now
+        else:
+            # Record new error
+            self.errors[key] = {
+                "type": error_info["type"],
+                "hook_type": hook_type,
+                "details": error_info["details"],
+                "match": error_info["match"],
+                "returncode": error_info.get("returncode", 1),
+                "count": 1,
+                "first_seen": now,
+                "last_seen": now,
+            }
+
+        self._save_errors()
+        self.logger.debug(
+            f"Recorded error: {error_info['type']} (count: {self.errors[key]['count']})"
+        )
+
+    def is_known_failing_hook(self, hook_type: str) -> Optional[Dict[str, Any]]:
+        """Check if a hook type is known to fail repeatedly.
+
+        WHY check for 2+ failures:
+        - Single failure could be transient
+        - 2+ failures indicate persistent issue
+        - Balance between retry attempts and error prevention
+
+        Args:
+            hook_type: Type of hook to check
+
+        Returns:
+            Error data if hook is known to fail, None otherwise
+        """
+        # Find any errors for this hook type with 2+ occurrences
+        for key, error_data in self.errors.items():
+            if error_data["hook_type"] == hook_type and error_data["count"] >= 2:
+                return error_data
+
+        return None
+
+    def should_skip_hook(self, hook_type: str, threshold: int = 2) -> bool:
+        """Determine if a hook should be skipped due to repeated failures.
+
+        Args:
+            hook_type: Type of hook to check
+            threshold: Minimum failure count to skip (default: 2)
+
+        Returns:
+            True if hook should be skipped
+        """
+        error_data = self.is_known_failing_hook(hook_type)
+        return error_data is not None and error_data["count"] >= threshold
+
+    def suggest_fix(self, error_info: Dict[str, str]) -> str:
+        """Suggest a fix for the detected error.
+
+        WHY provide suggestions:
+        - Users need actionable guidance
+        - Common errors have known solutions
+        - Reduces support burden
+        - Improves user experience
+
+        Args:
+            error_info: Error information from detect_error()
+
+        Returns:
+            Human-readable fix suggestion
+        """
+        error_type = error_info["type"]
+        details = error_info.get("details", "")
+
+        suggestions = {
+            "file_not_found": f"""File not found: {details}
+
+Possible fixes:
+1. Check if the file exists: ls -la {details}
+2. Verify the path is correct in your hook configuration
+3. If it's a script, ensure it's executable: chmod +x {details}
+4. Clear error memory to retry: rm {self.memory_file}
+""",
+            "command_not_found": f"""Command not found: {details}
+
+Possible fixes:
+1. Install the missing command
+2. Check if it's in your PATH: which {details}
+3. Update hook configuration to use absolute path
+4. Remove the hook if no longer needed
+""",
+            "permission_denied": f"""Permission denied: {details}
+
+Possible fixes:
+1. Check file permissions: ls -la {details}
+2. Make file executable: chmod +x {details}
+3. Run with appropriate privileges
+4. Check file ownership
+""",
+            "syntax_error": """Syntax error in hook configuration or script
+
+Possible fixes:
+1. Review hook configuration in .claude-mpm/config
+2. Check script syntax if using shell hooks
+3. Validate JSON configuration format
+4. Check for typos in hook definitions
+""",
+            "eval_error": f"""Error in hook execution: {details}
+
+Possible fixes:
+1. Review hook handler logs for details
+2. Check hook configuration syntax
+3. Verify all required dependencies are available
+4. Test hook handler manually: python {details}
+""",
+            "general_error": f"""Error during hook execution: {error_info.get("match", "Unknown error")}
+
+Possible fixes:
+1. Check logs for detailed error information
+2. Verify hook configuration is correct
+3. Ensure all dependencies are installed
+4. Clear error memory to retry: rm {self.memory_file}
+""",
+        }
+
+        return suggestions.get(
+            error_type, f"Unknown error type: {error_type}\n\nDetails: {details}"
+        )
+
+    def clear_errors(self, hook_type: Optional[str] = None):
+        """Clear error memory to allow retry of failed hooks.
+
+        Args:
+            hook_type: If specified, only clear errors for this hook type
+        """
+        if hook_type is None:
+            # Clear all errors
+            count = len(self.errors)
+            self.errors.clear()
+            self._save_errors()
+            self.logger.info(f"Cleared all {count} error records")
+        else:
+            # Clear errors for specific hook type
+            keys_to_remove = [
+                key
+                for key, data in self.errors.items()
+                if data["hook_type"] == hook_type
+            ]
+            for key in keys_to_remove:
+                del self.errors[key]
+            self._save_errors()
+            self.logger.info(
+                f"Cleared {len(keys_to_remove)} error records for {hook_type}"
+            )
+
+    def get_error_summary(self) -> Dict[str, Any]:
+        """Get summary of all recorded errors.
+
+        Returns:
+            Dictionary with error statistics and details
+        """
+        if not self.errors:
+            return {
+                "total_errors": 0,
+                "unique_errors": 0,
+                "errors_by_type": {},
+                "errors_by_hook": {},
+            }
+
+        errors_by_type = {}
+        errors_by_hook = {}
+
+        total_count = 0
+
+        for error_data in self.errors.values():
+            error_type = error_data["type"]
+            hook_type = error_data["hook_type"]
+            count = error_data["count"]
+
+            total_count += count
+
+            errors_by_type[error_type] = errors_by_type.get(error_type, 0) + count
+            errors_by_hook[hook_type] = errors_by_hook.get(hook_type, 0) + count
+
+        return {
+            "total_errors": total_count,
+            "unique_errors": len(self.errors),
+            "errors_by_type": errors_by_type,
+            "errors_by_hook": errors_by_hook,
+            "memory_file": str(self.memory_file),
+        }
+
+
+# Global instance
+_hook_error_memory: Optional[HookErrorMemory] = None
+
+
+def get_hook_error_memory(memory_file: Optional[Path] = None) -> HookErrorMemory:
+    """Get the global hook error memory instance.
+
+    Args:
+        memory_file: Optional custom memory file path
+
+    Returns:
+        HookErrorMemory instance
+    """
+    global _hook_error_memory
+    if _hook_error_memory is None:
+        _hook_error_memory = HookErrorMemory(memory_file)
+    return _hook_error_memory
+
+
+def clear_hook_errors(hook_type: Optional[str] = None):
+    """Convenience function to clear hook error memory.
+
+    Args:
+        hook_type: If specified, only clear errors for this hook type
+    """
+    memory = get_hook_error_memory()
+    memory.clear_errors(hook_type)