claude-mpm 4.1.2__py3-none-any.whl → 4.1.4__py3-none-any.whl

claude_mpm/hooks/claude_hooks/services/state_manager.py

@@ -0,0 +1,282 @@
+"""State management service for Claude hook handler.
+
+This service manages:
+- Agent delegation tracking
+- Git branch caching
+- Session state management
+- Cleanup of old entries
+"""
+
+import os
+import subprocess
+import time
+from collections import deque
+from datetime import datetime
+from typing import Optional
+
+# Import constants for configuration
+try:
+    from claude_mpm.core.constants import TimeoutConfig
+except ImportError:
+    # Fallback values if constants module not available
+    class TimeoutConfig:
+        QUICK_TIMEOUT = 2.0
+
+
+# Debug mode is enabled by default for better visibility into hook processing
+DEBUG = os.environ.get("CLAUDE_MPM_HOOK_DEBUG", "true").lower() != "false"
+
+
+class StateManagerService:
+    """Manages state for the Claude hook handler."""
+
+    def __init__(self):
+        """Initialize state management service."""
+        # Maximum sizes for tracking
+        self.MAX_DELEGATION_TRACKING = 200
+        self.MAX_PROMPT_TRACKING = 100
+        self.MAX_CACHE_AGE_SECONDS = 300
+        self.CLEANUP_INTERVAL_EVENTS = 100
+
+        # Agent delegation tracking
+        # Store recent Task delegations: session_id -> agent_type
+        self.active_delegations = {}
+        # Use deque to limit memory usage (keep last 100 delegations)
+        self.delegation_history = deque(maxlen=100)
+        # Store delegation request data for response correlation: session_id -> request_data
+        self.delegation_requests = {}
+
+        # Git branch cache (to avoid repeated subprocess calls)
+        self._git_branch_cache = {}
+        self._git_branch_cache_time = {}
+
+        # Store current user prompts for comprehensive response tracking
+        self.pending_prompts = {}  # session_id -> prompt data
+
+        # Track events for periodic cleanup
+        self.events_processed = 0
+        self.last_cleanup = time.time()
+
+    def track_delegation(
+        self, session_id: str, agent_type: str, request_data: Optional[dict] = None
+    ):
+        """Track a new agent delegation with optional request data for response correlation."""
+        if DEBUG:
+            import sys
+
+            print(
+                f"  - session_id: {session_id[:16] if session_id else 'None'}...",
+                file=sys.stderr,
+            )
+            print(f"  - agent_type: {agent_type}", file=sys.stderr)
+            print(f"  - request_data provided: {bool(request_data)}", file=sys.stderr)
+            print(
+                f"  - delegation_requests size before: {len(self.delegation_requests)}",
+                file=sys.stderr,
+            )
+
+        if session_id and agent_type and agent_type != "unknown":
+            self.active_delegations[session_id] = agent_type
+            key = f"{session_id}:{datetime.now().timestamp()}"
+            self.delegation_history.append((key, agent_type))
+
+            # Store request data for response tracking correlation
+            if request_data:
+                self.delegation_requests[session_id] = {
+                    "agent_type": agent_type,
+                    "request": request_data,
+                    "timestamp": datetime.now().isoformat(),
+                }
+                if DEBUG:
+                    import sys
+
+                    print(
+                        f"  - ✅ Stored in delegation_requests[{session_id[:16]}...]",
+                        file=sys.stderr,
+                    )
+                    print(
+                        f"  - delegation_requests size after: {len(self.delegation_requests)}",
+                        file=sys.stderr,
+                    )
+
+            # Clean up old delegations (older than 5 minutes)
+            cutoff_time = datetime.now().timestamp() - 300
+            keys_to_remove = []
+            for sid in list(self.active_delegations.keys()):
+                # Check if this is an old entry by looking in history
+                found_recent = False
+                for hist_key, _ in reversed(self.delegation_history):
+                    if hist_key.startswith(sid):
+                        _, timestamp = hist_key.split(":", 1)
+                        if float(timestamp) > cutoff_time:
+                            found_recent = True
+                            break
+                if not found_recent:
+                    keys_to_remove.append(sid)
+
+            for key in keys_to_remove:
+                if key in self.active_delegations:
+                    del self.active_delegations[key]
+                if key in self.delegation_requests:
+                    del self.delegation_requests[key]
+
+    def get_delegation_agent_type(self, session_id: str) -> str:
+        """Get the agent type for a session's active delegation."""
+        # First try exact session match
+        if session_id and session_id in self.active_delegations:
+            return self.active_delegations[session_id]
+
+        # Then try to find in recent history
+        if session_id:
+            for key, agent_type in reversed(self.delegation_history):
+                if key.startswith(session_id):
+                    return agent_type
+
+        return "unknown"
+
+    def cleanup_old_entries(self):
+        """Clean up old entries to prevent memory growth."""
+        datetime.now().timestamp() - self.MAX_CACHE_AGE_SECONDS
+
+        # Clean up delegation tracking dictionaries
+        for storage in [self.active_delegations, self.delegation_requests]:
+            if len(storage) > self.MAX_DELEGATION_TRACKING:
+                # Keep only the most recent entries
+                sorted_keys = sorted(storage.keys())
+                excess = len(storage) - self.MAX_DELEGATION_TRACKING
+                for key in sorted_keys[:excess]:
+                    del storage[key]
+
+        # Clean up pending prompts
+        if len(self.pending_prompts) > self.MAX_PROMPT_TRACKING:
+            sorted_keys = sorted(self.pending_prompts.keys())
+            excess = len(self.pending_prompts) - self.MAX_PROMPT_TRACKING
+            for key in sorted_keys[:excess]:
+                del self.pending_prompts[key]
+
+        # Clean up git branch cache
+        expired_keys = [
+            key
+            for key, cache_time in self._git_branch_cache_time.items()
+            if datetime.now().timestamp() - cache_time > self.MAX_CACHE_AGE_SECONDS
+        ]
+        for key in expired_keys:
+            self._git_branch_cache.pop(key, None)
+            self._git_branch_cache_time.pop(key, None)
+
+    def get_git_branch(self, working_dir: Optional[str] = None) -> str:
+        """Get git branch for the given directory with caching.
+
+        WHY caching approach:
+        - Avoids repeated subprocess calls which are expensive
+        - Caches results for 30 seconds per directory
+        - Falls back gracefully if git command fails
+        - Returns 'Unknown' for non-git directories
+        """
+        # Use current working directory if not specified
+        if not working_dir:
+            working_dir = os.getcwd()
+
+        # Check cache first (cache for 30 seconds)
+        current_time = datetime.now().timestamp()
+        cache_key = working_dir
+
+        if (
+            cache_key in self._git_branch_cache
+            and cache_key in self._git_branch_cache_time
+            and current_time - self._git_branch_cache_time[cache_key] < 30
+        ):
+            return self._git_branch_cache[cache_key]
+
+        # Try to get git branch
+        try:
+            # Change to the working directory temporarily
+            original_cwd = os.getcwd()
+            os.chdir(working_dir)
+
+            # Run git command to get current branch
+            result = subprocess.run(
+                ["git", "branch", "--show-current"],
+                capture_output=True,
+                text=True,
+                timeout=TimeoutConfig.QUICK_TIMEOUT,
+                check=False,  # Quick timeout to avoid hanging
+            )
+
+            # Restore original directory
+            os.chdir(original_cwd)
+
+            if result.returncode == 0 and result.stdout.strip():
+                branch = result.stdout.strip()
+                # Cache the result
+                self._git_branch_cache[cache_key] = branch
+                self._git_branch_cache_time[cache_key] = current_time
+                return branch
+            # Not a git repository or no branch
+            self._git_branch_cache[cache_key] = "Unknown"
+            self._git_branch_cache_time[cache_key] = current_time
+            return "Unknown"
+
+        except (
+            subprocess.TimeoutExpired,
+            subprocess.CalledProcessError,
+            FileNotFoundError,
+            OSError,
+        ):
+            # Git not available or command failed
+            self._git_branch_cache[cache_key] = "Unknown"
+            self._git_branch_cache_time[cache_key] = current_time
+            return "Unknown"
+
+    def find_matching_request(self, session_id: str) -> Optional[dict]:
+        """Find matching request data for a session, with fuzzy matching fallback."""
+        # First try exact match
+        request_info = self.delegation_requests.get(session_id)
+
+        # If exact match fails, try partial matching
+        if not request_info and session_id:
+            if DEBUG:
+                import sys
+
+                print(
+                    f"  - Trying fuzzy match for session {session_id[:16]}...",
+                    file=sys.stderr,
+                )
+            # Try to find a session that matches the first 8-16 characters
+            for stored_sid in list(self.delegation_requests.keys()):
+                if (
+                    stored_sid.startswith(session_id[:8])
+                    or session_id.startswith(stored_sid[:8])
+                    or (
+                        len(session_id) >= 16
+                        and len(stored_sid) >= 16
+                        and stored_sid[:16] == session_id[:16]
+                    )
+                ):
+                    if DEBUG:
+                        import sys
+
+                        print(
+                            f"  - ✅ Fuzzy match found: {stored_sid[:16]}...",
+                            file=sys.stderr,
+                        )
+                    request_info = self.delegation_requests.get(stored_sid)
+                    # Update the key to use the current session_id for consistency
+                    if request_info:
+                        self.delegation_requests[session_id] = request_info
+                        # Optionally remove the old key to avoid duplicates
+                        if stored_sid != session_id:
+                            del self.delegation_requests[stored_sid]
+                    break
+
+        return request_info
+
+    def remove_request(self, session_id: str):
+        """Remove request data for a session."""
+        if session_id in self.delegation_requests:
+            del self.delegation_requests[session_id]
+
+    def increment_events_processed(self) -> bool:
+        """Increment events processed counter and return True if cleanup is needed."""
+        self.events_processed += 1
+        return self.events_processed % self.CLEANUP_INTERVAL_EVENTS == 0

claude_mpm/hooks/claude_hooks/services/subagent_processor.py

@@ -0,0 +1,374 @@
+"""Subagent response processing service for Claude hook handler.
+
+This service handles:
+- SubagentStop event processing
+- Structured response extraction
+- Response tracking and correlation
+- Memory field processing
+"""
+
+import json
+import os
+import re
+import sys
+from datetime import datetime
+from typing import Optional, Tuple
+
+# Debug mode is enabled by default for better visibility into hook processing
+DEBUG = os.environ.get("CLAUDE_MPM_HOOK_DEBUG", "true").lower() != "false"
+
+
+class SubagentResponseProcessor:
+    """Processes subagent responses and extracts structured data."""
+
+    def __init__(self, state_manager, response_tracking_manager, connection_manager):
+        """Initialize the subagent response processor.
+
+        Args:
+            state_manager: StateManagerService instance
+            response_tracking_manager: ResponseTrackingManager instance
+            connection_manager: ConnectionManagerService instance
+        """
+        self.state_manager = state_manager
+        self.response_tracking_manager = response_tracking_manager
+        self.connection_manager = connection_manager
+
+    def process_subagent_stop(self, event: dict):
+        """Handle subagent stop events with improved agent type detection.
+
+        WHY comprehensive subagent stop capture:
+        - Provides visibility into subagent lifecycle and delegation patterns
+        - Captures agent type, ID, reason, and results for analysis
+        - Enables tracking of delegation success/failure patterns
+        - Useful for understanding subagent performance and reliability
+        """
+        # Enhanced debug logging for session correlation
+        session_id = event.get("session_id", "")
+        if DEBUG:
+            print(
+                f"  - session_id: {session_id[:16] if session_id else 'None'}...",
+                file=sys.stderr,
+            )
+            print(f"  - event keys: {list(event.keys())}", file=sys.stderr)
+            print(
+                f"  - delegation_requests size: {len(self.state_manager.delegation_requests)}",
+                file=sys.stderr,
+            )
+            # Show all stored session IDs for comparison
+            all_sessions = list(self.state_manager.delegation_requests.keys())
+            if all_sessions:
+                print("  - Stored sessions (first 16 chars):", file=sys.stderr)
+                for sid in all_sessions[:10]:  # Show up to 10
+                    print(
+                        f"    - {sid[:16]}... (agent: {self.state_manager.delegation_requests[sid].get('agent_type', 'unknown')})",
+                        file=sys.stderr,
+                    )
+            else:
+                print("  - No stored sessions in delegation_requests!", file=sys.stderr)
+
+        # Get agent type and other basic info
+        agent_type, agent_id, reason = self._extract_basic_info(event, session_id)
+
+        # Always log SubagentStop events for debugging
+        if DEBUG or agent_type != "unknown":
+            print(
+                f"Hook handler: Processing SubagentStop - agent: '{agent_type}', session: '{session_id}', reason: '{reason}'",
+                file=sys.stderr,
+            )
+
+        # Get working directory and git branch
+        working_dir = event.get("cwd", "")
+        git_branch = (
+            self.state_manager.get_git_branch(working_dir) if working_dir else "Unknown"
+        )
+
+        # Try to extract structured response from output if available
+        output = event.get("output", "")
+        structured_response = self._extract_structured_response(output, agent_type)
+
+        # Track agent response
+        self._track_response(
+            event,
+            session_id,
+            agent_type,
+            reason,
+            working_dir,
+            git_branch,
+            output,
+            structured_response,
+        )
+
+        # Build subagent stop data for event emission
+        subagent_stop_data = self._build_subagent_stop_data(
+            event,
+            session_id,
+            agent_type,
+            agent_id,
+            reason,
+            working_dir,
+            git_branch,
+            structured_response,
+        )
+
+        # Debug log the processed data
+        if DEBUG:
+            print(
+                f"SubagentStop processed data: agent_type='{agent_type}', session_id='{session_id}'",
+                file=sys.stderr,
+            )
+
+        # Emit to /hook namespace with high priority
+        self.connection_manager.emit_event("/hook", "subagent_stop", subagent_stop_data)
+
+    def _extract_basic_info(self, event: dict, session_id: str) -> Tuple[str, str, str]:
+        """Extract basic info from the event."""
+        # First try to get agent type from our tracking
+        agent_type = (
+            self.state_manager.get_delegation_agent_type(session_id)
+            if session_id
+            else "unknown"
+        )
+
+        # Fall back to event data if tracking didn't have it
+        if agent_type == "unknown":
+            agent_type = event.get("agent_type", event.get("subagent_type", "unknown"))
+
+        agent_id = event.get("agent_id", event.get("subagent_id", ""))
+        reason = event.get("reason", event.get("stop_reason", "unknown"))
+
+        # Try to infer agent type from other fields if still unknown
+        if agent_type == "unknown" and "task" in event:
+            task_desc = str(event.get("task", "")).lower()
+            if "research" in task_desc:
+                agent_type = "research"
+            elif "engineer" in task_desc or "code" in task_desc:
+                agent_type = "engineer"
+            elif "pm" in task_desc or "project" in task_desc:
+                agent_type = "pm"
+
+        return agent_type, agent_id, reason
+
+    def _extract_structured_response(
+        self, output: str, agent_type: str
+    ) -> Optional[dict]:
+        """Extract structured JSON response from output."""
+        if not output:
+            return None
+
+        try:
+            json_match = re.search(r"```json\s*(\{.*?\})\s*```", str(output), re.DOTALL)
+            if json_match:
+                structured_response = json.loads(json_match.group(1))
+                if DEBUG:
+                    print(
+                        f"Extracted structured response from {agent_type} agent in SubagentStop",
+                        file=sys.stderr,
+                    )
+
+                # Log if MEMORIES field is present
+                if structured_response.get("MEMORIES") and DEBUG:
+                    memories_count = len(structured_response["MEMORIES"])
+                    print(
+                        f"Agent {agent_type} returned MEMORIES field with {memories_count} items",
+                        file=sys.stderr,
+                    )
+
+                return structured_response
+        except (json.JSONDecodeError, AttributeError):
+            pass  # No structured response, that's okay
+
+        return None
+
+    def _track_response(
+        self,
+        event: dict,
+        session_id: str,
+        agent_type: str,
+        reason: str,
+        working_dir: str,
+        git_branch: str,
+        output: str,
+        structured_response: Optional[dict],
+    ):
+        """Track the agent response if response tracking is enabled."""
+        if DEBUG:
+            print(
+                f"  - response_tracking_enabled: {self.response_tracking_manager.response_tracking_enabled}",
+                file=sys.stderr,
+            )
+            print(
+                f"  - response_tracker exists: {self.response_tracking_manager.response_tracker is not None}",
+                file=sys.stderr,
+            )
+            print(
+                f"  - session_id: {session_id[:16] if session_id else 'None'}...",
+                file=sys.stderr,
+            )
+            print(f"  - agent_type: {agent_type}", file=sys.stderr)
+            print(f"  - reason: {reason}", file=sys.stderr)
+
+        if (
+            self.response_tracking_manager.response_tracking_enabled
+            and self.response_tracking_manager.response_tracker
+        ):
+            try:
+                # Get the original request data (with fuzzy matching fallback)
+                request_info = self.state_manager.find_matching_request(session_id)
+
+                if DEBUG:
+                    print(
+                        f"  - request_info present: {bool(request_info)}",
+                        file=sys.stderr,
+                    )
+                    if request_info:
+                        print(
+                            "  - ✅ Found request data for response tracking",
+                            file=sys.stderr,
+                        )
+                        print(
+                            f"  - stored agent_type: {request_info.get('agent_type')}",
+                            file=sys.stderr,
+                        )
+                        print(
+                            f"  - request keys: {list(request_info.get('request', {}).keys())}",
+                            file=sys.stderr,
+                        )
+                    else:
+                        print(
+                            f"  - ❌ No request data found for session {session_id[:16]}...",
+                            file=sys.stderr,
+                        )
+
+                if request_info:
+                    # Use the output as the response
+                    response_text = (
+                        str(output)
+                        if output
+                        else f"Agent {agent_type} completed with reason: {reason}"
+                    )
+
+                    # Get the original request
+                    original_request = request_info.get("request", {})
+                    prompt = original_request.get("prompt", "")
+                    description = original_request.get("description", "")
+
+                    # Combine prompt and description
+                    full_request = prompt
+                    if description and description != prompt:
+                        if full_request:
+                            full_request += f"\n\nDescription: {description}"
+                        else:
+                            full_request = description
+
+                    if not full_request:
+                        full_request = f"Task delegation to {agent_type} agent"
+
+                    # Prepare metadata
+                    metadata = {
+                        "exit_code": event.get("exit_code", 0),
+                        "success": reason in ["completed", "finished", "done"],
+                        "has_error": reason
+                        in ["error", "timeout", "failed", "blocked"],
+                        "duration_ms": event.get("duration_ms"),
+                        "working_directory": working_dir,
+                        "git_branch": git_branch,
+                        "timestamp": datetime.now().isoformat(),
+                        "event_type": "subagent_stop",
+                        "reason": reason,
+                        "original_request_timestamp": request_info.get("timestamp"),
+                    }
+
+                    # Add structured response if available
+                    if structured_response:
+                        metadata["structured_response"] = structured_response
+                        metadata["task_completed"] = structured_response.get(
+                            "task_completed", False
+                        )
+
+                        # Check for MEMORIES field and process if present
+                        if structured_response.get("MEMORIES") and DEBUG:
+                            memories = structured_response["MEMORIES"]
+                            print(
+                                f"Found MEMORIES field in {agent_type} response with {len(memories)} items",
+                                file=sys.stderr,
+                            )
+                            # The memory will be processed by extract_and_update_memory
+                            # which is called by the memory hook service
+
+                    # Track the response
+                    file_path = (
+                        self.response_tracking_manager.response_tracker.track_response(
+                            agent_name=agent_type,
+                            request=full_request,
+                            response=response_text,
+                            session_id=session_id,
+                            metadata=metadata,
+                        )
+                    )
+
+                    if file_path and DEBUG:
+                        print(
+                            f"✅ Tracked {agent_type} agent response on SubagentStop: {file_path.name}",
+                            file=sys.stderr,
+                        )
+
+                    # Clean up the request data
+                    self.state_manager.remove_request(session_id)
+
+                elif DEBUG:
+                    print(
+                        f"No request data for SubagentStop session {session_id[:8]}..., agent: {agent_type}",
+                        file=sys.stderr,
+                    )
+
+            except Exception as e:
+                if DEBUG:
+                    print(
+                        f"❌ Failed to track response on SubagentStop: {e}",
+                        file=sys.stderr,
+                    )
+
+    def _build_subagent_stop_data(
+        self,
+        event: dict,
+        session_id: str,
+        agent_type: str,
+        agent_id: str,
+        reason: str,
+        working_dir: str,
+        git_branch: str,
+        structured_response: Optional[dict],
+    ) -> dict:
+        """Build the subagent stop data for event emission."""
+        subagent_stop_data = {
+            "agent_type": agent_type,
+            "agent_id": agent_id,
+            "reason": reason,
+            "session_id": session_id,
+            "working_directory": working_dir,
+            "git_branch": git_branch,
+            "timestamp": datetime.now().isoformat(),
+            "is_successful_completion": reason in ["completed", "finished", "done"],
+            "is_error_termination": reason in ["error", "timeout", "failed", "blocked"],
+            "is_delegation_related": agent_type
+            in ["research", "engineer", "pm", "ops", "qa", "documentation", "security"],
+            "has_results": bool(event.get("results") or event.get("output")),
+            "duration_context": event.get("duration_ms"),
+            "hook_event_name": "SubagentStop",  # Explicitly set for dashboard
+        }
+
+        # Add structured response data if available
+        if structured_response:
+            subagent_stop_data["structured_response"] = {
+                "task_completed": structured_response.get("task_completed", False),
+                "instructions": structured_response.get("instructions", ""),
+                "results": structured_response.get("results", ""),
+                "files_modified": structured_response.get("files_modified", []),
+                "tools_used": structured_response.get("tools_used", []),
+                "remember": structured_response.get("remember"),
+                "MEMORIES": structured_response.get(
+                    "MEMORIES"
+                ),  # Complete memory replacement
+            }
+
+        return subagent_stop_data