alma-memory 0.5.0__py3-none-any.whl → 0.7.0__py3-none-any.whl

alma/storage/file_based.py

@@ -1,619 +1,614 @@
-"""
-ALMA File-Based Storage Backend.
-
-Simple JSON file storage for testing and fallback scenarios.
-No vector search - uses basic text matching for retrieval.
-"""
-
-import json
-import logging
-from datetime import datetime, timezone
-from pathlib import Path
-from typing import Any, Dict, List, Optional
-
-from alma.storage.base import StorageBackend
-from alma.types import (
-    AntiPattern,
-    DomainKnowledge,
-    Heuristic,
-    Outcome,
-    UserPreference,
-)
-
-logger = logging.getLogger(__name__)
-
-
-class FileBasedStorage(StorageBackend):
-    """
-    File-based storage using JSON files.
-
-    Structure:
-        .alma/
-        ├── heuristics.json
-        ├── outcomes.json
-        ├── preferences.json
-        ├── domain_knowledge.json
-        └── anti_patterns.json
-
-    Note: This backend does NOT support vector search.
-    Use SQLiteStorage or AzureCosmosStorage for semantic retrieval.
-    """
-
-    def __init__(self, storage_dir: Path):
-        """
-        Initialize file-based storage.
-
-        Args:
-            storage_dir: Directory to store JSON files
-        """
-        self.storage_dir = Path(storage_dir)
-        self.storage_dir.mkdir(parents=True, exist_ok=True)
-
-        # File paths
-        self._files = {
-            "heuristics": self.storage_dir / "heuristics.json",
-            "outcomes": self.storage_dir / "outcomes.json",
-            "preferences": self.storage_dir / "preferences.json",
-            "domain_knowledge": self.storage_dir / "domain_knowledge.json",
-            "anti_patterns": self.storage_dir / "anti_patterns.json",
-        }
-
-        # Initialize empty files if they don't exist
-        for file_path in self._files.values():
-            if not file_path.exists():
-                self._write_json(file_path, [])
-
-    @classmethod
-    def from_config(cls, config: Dict[str, Any]) -> "FileBasedStorage":
-        """Create instance from configuration."""
-        storage_dir = config.get("storage_dir", ".alma")
-        return cls(storage_dir=Path(storage_dir))
-
-    # ==================== WRITE OPERATIONS ====================
-
-    def save_heuristic(self, heuristic: Heuristic) -> str:
-        """Save a heuristic (UPSERT - update if exists, insert if new)."""
-        data = self._read_json(self._files["heuristics"])
-        record = self._to_dict(heuristic)
-        # Find and replace existing, or append new
-        found = False
-        for i, existing in enumerate(data):
-            if existing.get("id") == record["id"]:
-                data[i] = record
-                found = True
-                break
-        if not found:
-            data.append(record)
-        self._write_json(self._files["heuristics"], data)
-        logger.debug(f"Saved heuristic: {heuristic.id}")
-        return heuristic.id
-
-    def save_outcome(self, outcome: Outcome) -> str:
-        """Save an outcome (UPSERT - update if exists, insert if new)."""
-        data = self._read_json(self._files["outcomes"])
-        record = self._to_dict(outcome)
-        # Find and replace existing, or append new
-        found = False
-        for i, existing in enumerate(data):
-            if existing.get("id") == record["id"]:
-                data[i] = record
-                found = True
-                break
-        if not found:
-            data.append(record)
-        self._write_json(self._files["outcomes"], data)
-        logger.debug(f"Saved outcome: {outcome.id}")
-        return outcome.id
-
-    def save_user_preference(self, preference: UserPreference) -> str:
-        """Save a user preference (UPSERT - update if exists, insert if new)."""
-        data = self._read_json(self._files["preferences"])
-        record = self._to_dict(preference)
-        # Find and replace existing, or append new
-        found = False
-        for i, existing in enumerate(data):
-            if existing.get("id") == record["id"]:
-                data[i] = record
-                found = True
-                break
-        if not found:
-            data.append(record)
-        self._write_json(self._files["preferences"], data)
-        logger.debug(f"Saved preference: {preference.id}")
-        return preference.id
-
-    def save_domain_knowledge(self, knowledge: DomainKnowledge) -> str:
-        """Save domain knowledge (UPSERT - update if exists, insert if new)."""
-        data = self._read_json(self._files["domain_knowledge"])
-        record = self._to_dict(knowledge)
-        # Find and replace existing, or append new
-        found = False
-        for i, existing in enumerate(data):
-            if existing.get("id") == record["id"]:
-                data[i] = record
-                found = True
-                break
-        if not found:
-            data.append(record)
-        self._write_json(self._files["domain_knowledge"], data)
-        logger.debug(f"Saved domain knowledge: {knowledge.id}")
-        return knowledge.id
-
-    def save_anti_pattern(self, anti_pattern: AntiPattern) -> str:
-        """Save an anti-pattern (UPSERT - update if exists, insert if new)."""
-        data = self._read_json(self._files["anti_patterns"])
-        record = self._to_dict(anti_pattern)
-        # Find and replace existing, or append new
-        found = False
-        for i, existing in enumerate(data):
-            if existing.get("id") == record["id"]:
-                data[i] = record
-                found = True
-                break
-        if not found:
-            data.append(record)
-        self._write_json(self._files["anti_patterns"], data)
-        logger.debug(f"Saved anti-pattern: {anti_pattern.id}")
-        return anti_pattern.id
-
-    # ==================== READ OPERATIONS ====================
-
-    def get_heuristics(
-        self,
-        project_id: str,
-        agent: Optional[str] = None,
-        embedding: Optional[List[float]] = None,
-        top_k: int = 5,
-        min_confidence: float = 0.0,
-    ) -> List[Heuristic]:
-        """Get heuristics (no vector search - returns all matching filters)."""
-        data = self._read_json(self._files["heuristics"])
-
-        # Filter
-        results = []
-        for record in data:
-            if record.get("project_id") != project_id:
-                continue
-            if agent and record.get("agent") != agent:
-                continue
-            if record.get("confidence", 0) < min_confidence:
-                continue
-            results.append(self._to_heuristic(record))
-
-        # Sort by confidence and return top_k
-        results.sort(key=lambda x: -x.confidence)
-        return results[:top_k]
-
-    def get_outcomes(
-        self,
-        project_id: str,
-        agent: Optional[str] = None,
-        task_type: Optional[str] = None,
-        embedding: Optional[List[float]] = None,
-        top_k: int = 5,
-        success_only: bool = False,
-    ) -> List[Outcome]:
-        """Get outcomes (no vector search)."""
-        data = self._read_json(self._files["outcomes"])
-
-        results = []
-        for record in data:
-            if record.get("project_id") != project_id:
-                continue
-            if agent and record.get("agent") != agent:
-                continue
-            if task_type and record.get("task_type") != task_type:
-                continue
-            if success_only and not record.get("success"):
-                continue
-            results.append(self._to_outcome(record))
-
-        # Sort by timestamp (most recent first) and return top_k
-        results.sort(key=lambda x: x.timestamp, reverse=True)
-        return results[:top_k]
-
-    def get_user_preferences(
-        self,
-        user_id: str,
-        category: Optional[str] = None,
-    ) -> List[UserPreference]:
-        """Get user preferences."""
-        data = self._read_json(self._files["preferences"])
-
-        results = []
-        for record in data:
-            if record.get("user_id") != user_id:
-                continue
-            if category and record.get("category") != category:
-                continue
-            results.append(self._to_user_preference(record))
-
-        return results
-
-    def get_domain_knowledge(
-        self,
-        project_id: str,
-        agent: Optional[str] = None,
-        domain: Optional[str] = None,
-        embedding: Optional[List[float]] = None,
-        top_k: int = 5,
-    ) -> List[DomainKnowledge]:
-        """Get domain knowledge (no vector search)."""
-        data = self._read_json(self._files["domain_knowledge"])
-
-        results = []
-        for record in data:
-            if record.get("project_id") != project_id:
-                continue
-            if agent and record.get("agent") != agent:
-                continue
-            if domain and record.get("domain") != domain:
-                continue
-            results.append(self._to_domain_knowledge(record))
-
-        # Sort by confidence and return top_k
-        results.sort(key=lambda x: -x.confidence)
-        return results[:top_k]
-
-    def get_anti_patterns(
-        self,
-        project_id: str,
-        agent: Optional[str] = None,
-        embedding: Optional[List[float]] = None,
-        top_k: int = 5,
-    ) -> List[AntiPattern]:
-        """Get anti-patterns (no vector search)."""
-        data = self._read_json(self._files["anti_patterns"])
-
-        results = []
-        for record in data:
-            if record.get("project_id") != project_id:
-                continue
-            if agent and record.get("agent") != agent:
-                continue
-            results.append(self._to_anti_pattern(record))
-
-        # Sort by occurrence count and return top_k
-        results.sort(key=lambda x: -x.occurrence_count)
-        return results[:top_k]
-
-    # ==================== UPDATE OPERATIONS ====================
-
-    def update_heuristic(
-        self,
-        heuristic_id: str,
-        updates: Dict[str, Any],
-    ) -> bool:
-        """Update a heuristic's fields."""
-        data = self._read_json(self._files["heuristics"])
-
-        for i, record in enumerate(data):
-            if record.get("id") == heuristic_id:
-                data[i].update(updates)
-                self._write_json(self._files["heuristics"], data)
-                return True
-
-        return False
-
-    def increment_heuristic_occurrence(
-        self,
-        heuristic_id: str,
-        success: bool,
-    ) -> bool:
-        """Increment heuristic occurrence count."""
-        data = self._read_json(self._files["heuristics"])
-
-        for i, record in enumerate(data):
-            if record.get("id") == heuristic_id:
-                data[i]["occurrence_count"] = record.get("occurrence_count", 0) + 1
-                if success:
-                    data[i]["success_count"] = record.get("success_count", 0) + 1
-                data[i]["last_validated"] = datetime.now(timezone.utc).isoformat()
-                self._write_json(self._files["heuristics"], data)
-                return True
-
-        return False
-
-    # ==================== UPDATE CONFIDENCE OPERATIONS ====================
-
-    def update_heuristic_confidence(
-        self,
-        heuristic_id: str,
-        new_confidence: float,
-    ) -> bool:
-        """Update a heuristic's confidence score."""
-        data = self._read_json(self._files["heuristics"])
-
-        for i, record in enumerate(data):
-            if record.get("id") == heuristic_id:
-                data[i]["confidence"] = new_confidence
-                data[i]["last_validated"] = datetime.now(timezone.utc).isoformat()
-                self._write_json(self._files["heuristics"], data)
-                return True
-
-        return False
-
-    def update_knowledge_confidence(
-        self,
-        knowledge_id: str,
-        new_confidence: float,
-    ) -> bool:
-        """Update domain knowledge confidence score."""
-        data = self._read_json(self._files["domain_knowledge"])
-
-        for i, record in enumerate(data):
-            if record.get("id") == knowledge_id:
-                data[i]["confidence"] = new_confidence
-                data[i]["last_verified"] = datetime.now(timezone.utc).isoformat()
-                self._write_json(self._files["domain_knowledge"], data)
-                return True
-
-        return False
-
-    # ==================== DELETE OPERATIONS ====================
-
-    def delete_heuristic(self, heuristic_id: str) -> bool:
-        """Delete a single heuristic by ID."""
-        data = self._read_json(self._files["heuristics"])
-        original_count = len(data)
-
-        filtered = [r for r in data if r.get("id") != heuristic_id]
-        self._write_json(self._files["heuristics"], filtered)
-
-        deleted = original_count != len(filtered)
-        if deleted:
-            logger.debug(f"Deleted heuristic: {heuristic_id}")
-        return deleted
-
-    def delete_outcome(self, outcome_id: str) -> bool:
-        """Delete a single outcome by ID."""
-        data = self._read_json(self._files["outcomes"])
-        original_count = len(data)
-
-        filtered = [r for r in data if r.get("id") != outcome_id]
-        self._write_json(self._files["outcomes"], filtered)
-
-        deleted = original_count != len(filtered)
-        if deleted:
-            logger.debug(f"Deleted outcome: {outcome_id}")
-        return deleted
-
-    def delete_domain_knowledge(self, knowledge_id: str) -> bool:
-        """Delete a single domain knowledge entry by ID."""
-        data = self._read_json(self._files["domain_knowledge"])
-        original_count = len(data)
-
-        filtered = [r for r in data if r.get("id") != knowledge_id]
-        self._write_json(self._files["domain_knowledge"], filtered)
-
-        deleted = original_count != len(filtered)
-        if deleted:
-            logger.debug(f"Deleted domain knowledge: {knowledge_id}")
-        return deleted
-
-    def delete_anti_pattern(self, anti_pattern_id: str) -> bool:
-        """Delete a single anti-pattern by ID."""
-        data = self._read_json(self._files["anti_patterns"])
-        original_count = len(data)
-
-        filtered = [r for r in data if r.get("id") != anti_pattern_id]
-        self._write_json(self._files["anti_patterns"], filtered)
-
-        deleted = original_count != len(filtered)
-        if deleted:
-            logger.debug(f"Deleted anti-pattern: {anti_pattern_id}")
-        return deleted
-
-    def delete_outcomes_older_than(
-        self,
-        project_id: str,
-        older_than: datetime,
-        agent: Optional[str] = None,
-    ) -> int:
-        """Delete old outcomes."""
-        data = self._read_json(self._files["outcomes"])
-        original_count = len(data)
-
-        filtered = []
-        for record in data:
-            if record.get("project_id") != project_id:
-                filtered.append(record)
-                continue
-            if agent and record.get("agent") != agent:
-                filtered.append(record)
-                continue
-
-            timestamp = self._parse_datetime(record.get("timestamp"))
-            if timestamp and timestamp >= older_than:
-                filtered.append(record)
-
-        self._write_json(self._files["outcomes"], filtered)
-        deleted = original_count - len(filtered)
-        logger.info(f"Deleted {deleted} old outcomes")
-        return deleted
-
-    def delete_low_confidence_heuristics(
-        self,
-        project_id: str,
-        below_confidence: float,
-        agent: Optional[str] = None,
-    ) -> int:
-        """Delete low-confidence heuristics."""
-        data = self._read_json(self._files["heuristics"])
-        original_count = len(data)
-
-        filtered = []
-        for record in data:
-            if record.get("project_id") != project_id:
-                filtered.append(record)
-                continue
-            if agent and record.get("agent") != agent:
-                filtered.append(record)
-                continue
-
-            if record.get("confidence", 0) >= below_confidence:
-                filtered.append(record)
-
-        self._write_json(self._files["heuristics"], filtered)
-        deleted = original_count - len(filtered)
-        logger.info(f"Deleted {deleted} low-confidence heuristics")
-        return deleted
-
-    # ==================== STATS ====================
-
-    def get_stats(
-        self,
-        project_id: str,
-        agent: Optional[str] = None,
-    ) -> Dict[str, Any]:
-        """Get memory statistics."""
-        stats = {
-            "project_id": project_id,
-            "agent": agent,
-            "heuristics_count": 0,
-            "outcomes_count": 0,
-            "preferences_count": 0,
-            "domain_knowledge_count": 0,
-            "anti_patterns_count": 0,
-        }
-
-        for name, file_path in self._files.items():
-            data = self._read_json(file_path)
-            count = 0
-            for record in data:
-                if name == "preferences":
-                    # Preferences don't have project_id
-                    count += 1
-                elif record.get("project_id") == project_id:
-                    if agent is None or record.get("agent") == agent:
-                        count += 1
-            stats[f"{name}_count"] = count
-
-        stats["total_count"] = sum(stats[k] for k in stats if k.endswith("_count"))
-
-        return stats
-
-    # ==================== HELPERS ====================
-
-    def _read_json(self, file_path: Path) -> List[Dict]:
-        """Read JSON file."""
-        try:
-            with open(file_path, "r") as f:
-                return json.load(f)
-        except (json.JSONDecodeError, FileNotFoundError):
-            return []
-
-    def _write_json(self, file_path: Path, data: List[Dict]):
-        """Write JSON file."""
-        with open(file_path, "w") as f:
-            json.dump(data, f, indent=2, default=str)
-
-    def _to_dict(self, obj: Any) -> Dict:
-        """Convert dataclass to dict with datetime handling."""
-        if hasattr(obj, "__dataclass_fields__"):
-            result = {}
-            for field_name in obj.__dataclass_fields__:
-                value = getattr(obj, field_name)
-                if isinstance(value, datetime):
-                    result[field_name] = value.isoformat()
-                elif value is not None:
-                    result[field_name] = value
-            return result
-        return dict(obj)
-
-    def _parse_datetime(self, value: Any) -> Optional[datetime]:
-        """Parse datetime from string or return as-is."""
-        if value is None:
-            return None
-        if isinstance(value, datetime):
-            return value
-        try:
-            return datetime.fromisoformat(value.replace("Z", "+00:00"))
-        except (ValueError, AttributeError):
-            return None
-
-    def _to_heuristic(self, record: Dict) -> Heuristic:
-        """Convert dict to Heuristic."""
-        return Heuristic(
-            id=record["id"],
-            agent=record["agent"],
-            project_id=record["project_id"],
-            condition=record["condition"],
-            strategy=record["strategy"],
-            confidence=record.get("confidence", 0.0),
-            occurrence_count=record.get("occurrence_count", 0),
-            success_count=record.get("success_count", 0),
-            last_validated=self._parse_datetime(record.get("last_validated"))
-            or datetime.now(timezone.utc),
-            created_at=self._parse_datetime(record.get("created_at"))
-            or datetime.now(timezone.utc),
-            embedding=record.get("embedding"),
-            metadata=record.get("metadata", {}),
-        )
-
-    def _to_outcome(self, record: Dict) -> Outcome:
-        """Convert dict to Outcome."""
-        return Outcome(
-            id=record["id"],
-            agent=record["agent"],
-            project_id=record["project_id"],
-            task_type=record.get("task_type", "general"),
-            task_description=record["task_description"],
-            success=record.get("success", False),
-            strategy_used=record.get("strategy_used", ""),
-            duration_ms=record.get("duration_ms"),
-            error_message=record.get("error_message"),
-            user_feedback=record.get("user_feedback"),
-            timestamp=self._parse_datetime(record.get("timestamp"))
-            or datetime.now(timezone.utc),
-            embedding=record.get("embedding"),
-            metadata=record.get("metadata", {}),
-        )
-
-    def _to_user_preference(self, record: Dict) -> UserPreference:
-        """Convert dict to UserPreference."""
-        return UserPreference(
-            id=record["id"],
-            user_id=record["user_id"],
-            category=record.get("category", "general"),
-            preference=record["preference"],
-            source=record.get("source", "unknown"),
-            confidence=record.get("confidence", 1.0),
-            timestamp=self._parse_datetime(record.get("timestamp"))
-            or datetime.now(timezone.utc),
-            metadata=record.get("metadata", {}),
-        )
-
-    def _to_domain_knowledge(self, record: Dict) -> DomainKnowledge:
-        """Convert dict to DomainKnowledge."""
-        return DomainKnowledge(
-            id=record["id"],
-            agent=record["agent"],
-            project_id=record["project_id"],
-            domain=record.get("domain", "general"),
-            fact=record["fact"],
-            source=record.get("source", "unknown"),
-            confidence=record.get("confidence", 1.0),
-            last_verified=self._parse_datetime(record.get("last_verified"))
-            or datetime.now(timezone.utc),
-            embedding=record.get("embedding"),
-            metadata=record.get("metadata", {}),
-        )
-
-    def _to_anti_pattern(self, record: Dict) -> AntiPattern:
-        """Convert dict to AntiPattern."""
-        return AntiPattern(
-            id=record["id"],
-            agent=record["agent"],
-            project_id=record["project_id"],
-            pattern=record["pattern"],
-            why_bad=record.get("why_bad", ""),
-            better_alternative=record.get("better_alternative", ""),
-            occurrence_count=record.get("occurrence_count", 1),
-            last_seen=self._parse_datetime(record.get("last_seen"))
-            or datetime.now(timezone.utc),
-            created_at=self._parse_datetime(record.get("created_at"))
-            or datetime.now(timezone.utc),
-            embedding=record.get("embedding"),
-            metadata=record.get("metadata", {}),
-        )
+"""
+ALMA File-Based Storage Backend.
+
+Simple JSON file storage for testing and fallback scenarios.
+No vector search - uses basic text matching for retrieval.
+"""
+
+import json
+import logging
+from datetime import datetime, timezone
+from pathlib import Path
+from typing import Any, Dict, List, Optional
+
+from alma.storage.base import StorageBackend
+from alma.storage.constants import MemoryType
+from alma.types import (
+    AntiPattern,
+    DomainKnowledge,
+    Heuristic,
+    Outcome,
+    UserPreference,
+)
+
+logger = logging.getLogger(__name__)
+
+
+class FileBasedStorage(StorageBackend):
+    """
+    File-based storage using JSON files.
+
+    Structure:
+        .alma/
+        ├── heuristics.json
+        ├── outcomes.json
+        ├── preferences.json
+        ├── domain_knowledge.json
+        └── anti_patterns.json
+
+    Note: This backend does NOT support vector search.
+    Use SQLiteStorage or AzureCosmosStorage for semantic retrieval.
+    """
+
+    def __init__(self, storage_dir: Path):
+        """
+        Initialize file-based storage.
+
+        Args:
+            storage_dir: Directory to store JSON files
+        """
+        self.storage_dir = Path(storage_dir)
+        self.storage_dir.mkdir(parents=True, exist_ok=True)
+
+        # File paths (using canonical memory type names)
+        self._files = {mt: self.storage_dir / f"{mt}.json" for mt in MemoryType.ALL}
+
+        # Initialize empty files if they don't exist
+        for file_path in self._files.values():
+            if not file_path.exists():
+                self._write_json(file_path, [])
+
+    @classmethod
+    def from_config(cls, config: Dict[str, Any]) -> "FileBasedStorage":
+        """Create instance from configuration."""
+        storage_dir = config.get("storage_dir", ".alma")
+        return cls(storage_dir=Path(storage_dir))
+
+    # ==================== WRITE OPERATIONS ====================
+
+    def save_heuristic(self, heuristic: Heuristic) -> str:
+        """Save a heuristic (UPSERT - update if exists, insert if new)."""
+        data = self._read_json(self._files["heuristics"])
+        record = self._to_dict(heuristic)
+        # Find and replace existing, or append new
+        found = False
+        for i, existing in enumerate(data):
+            if existing.get("id") == record["id"]:
+                data[i] = record
+                found = True
+                break
+        if not found:
+            data.append(record)
+        self._write_json(self._files["heuristics"], data)
+        logger.debug(f"Saved heuristic: {heuristic.id}")
+        return heuristic.id
+
+    def save_outcome(self, outcome: Outcome) -> str:
+        """Save an outcome (UPSERT - update if exists, insert if new)."""
+        data = self._read_json(self._files["outcomes"])
+        record = self._to_dict(outcome)
+        # Find and replace existing, or append new
+        found = False
+        for i, existing in enumerate(data):
+            if existing.get("id") == record["id"]:
+                data[i] = record
+                found = True
+                break
+        if not found:
+            data.append(record)
+        self._write_json(self._files["outcomes"], data)
+        logger.debug(f"Saved outcome: {outcome.id}")
+        return outcome.id
+
+    def save_user_preference(self, preference: UserPreference) -> str:
+        """Save a user preference (UPSERT - update if exists, insert if new)."""
+        data = self._read_json(self._files["preferences"])
+        record = self._to_dict(preference)
+        # Find and replace existing, or append new
+        found = False
+        for i, existing in enumerate(data):
+            if existing.get("id") == record["id"]:
+                data[i] = record
+                found = True
+                break
+        if not found:
+            data.append(record)
+        self._write_json(self._files["preferences"], data)
+        logger.debug(f"Saved preference: {preference.id}")
+        return preference.id
+
+    def save_domain_knowledge(self, knowledge: DomainKnowledge) -> str:
+        """Save domain knowledge (UPSERT - update if exists, insert if new)."""
+        data = self._read_json(self._files["domain_knowledge"])
+        record = self._to_dict(knowledge)
+        # Find and replace existing, or append new
+        found = False
+        for i, existing in enumerate(data):
+            if existing.get("id") == record["id"]:
+                data[i] = record
+                found = True
+                break
+        if not found:
+            data.append(record)
+        self._write_json(self._files["domain_knowledge"], data)
+        logger.debug(f"Saved domain knowledge: {knowledge.id}")
+        return knowledge.id
+
+    def save_anti_pattern(self, anti_pattern: AntiPattern) -> str:
+        """Save an anti-pattern (UPSERT - update if exists, insert if new)."""
+        data = self._read_json(self._files["anti_patterns"])
+        record = self._to_dict(anti_pattern)
+        # Find and replace existing, or append new
+        found = False
+        for i, existing in enumerate(data):
+            if existing.get("id") == record["id"]:
+                data[i] = record
+                found = True
+                break
+        if not found:
+            data.append(record)
+        self._write_json(self._files["anti_patterns"], data)
+        logger.debug(f"Saved anti-pattern: {anti_pattern.id}")
+        return anti_pattern.id
+
+    # ==================== READ OPERATIONS ====================
+
+    def get_heuristics(
+        self,
+        project_id: str,
+        agent: Optional[str] = None,
+        embedding: Optional[List[float]] = None,
+        top_k: int = 5,
+        min_confidence: float = 0.0,
+    ) -> List[Heuristic]:
+        """Get heuristics (no vector search - returns all matching filters)."""
+        data = self._read_json(self._files["heuristics"])
+
+        # Filter
+        results = []
+        for record in data:
+            if record.get("project_id") != project_id:
+                continue
+            if agent and record.get("agent") != agent:
+                continue
+            if record.get("confidence", 0) < min_confidence:
+                continue
+            results.append(self._to_heuristic(record))
+
+        # Sort by confidence and return top_k
+        results.sort(key=lambda x: -x.confidence)
+        return results[:top_k]
+
+    def get_outcomes(
+        self,
+        project_id: str,
+        agent: Optional[str] = None,
+        task_type: Optional[str] = None,
+        embedding: Optional[List[float]] = None,
+        top_k: int = 5,
+        success_only: bool = False,
+    ) -> List[Outcome]:
+        """Get outcomes (no vector search)."""
+        data = self._read_json(self._files["outcomes"])
+
+        results = []
+        for record in data:
+            if record.get("project_id") != project_id:
+                continue
+            if agent and record.get("agent") != agent:
+                continue
+            if task_type and record.get("task_type") != task_type:
+                continue
+            if success_only and not record.get("success"):
+                continue
+            results.append(self._to_outcome(record))
+
+        # Sort by timestamp (most recent first) and return top_k
+        results.sort(key=lambda x: x.timestamp, reverse=True)
+        return results[:top_k]
+
+    def get_user_preferences(
+        self,
+        user_id: str,
+        category: Optional[str] = None,
+    ) -> List[UserPreference]:
+        """Get user preferences."""
+        data = self._read_json(self._files["preferences"])
+
+        results = []
+        for record in data:
+            if record.get("user_id") != user_id:
+                continue
+            if category and record.get("category") != category:
+                continue
+            results.append(self._to_user_preference(record))
+
+        return results
+
+    def get_domain_knowledge(
+        self,
+        project_id: str,
+        agent: Optional[str] = None,
+        domain: Optional[str] = None,
+        embedding: Optional[List[float]] = None,
+        top_k: int = 5,
+    ) -> List[DomainKnowledge]:
+        """Get domain knowledge (no vector search)."""
+        data = self._read_json(self._files["domain_knowledge"])
+
+        results = []
+        for record in data:
+            if record.get("project_id") != project_id:
+                continue
+            if agent and record.get("agent") != agent:
+                continue
+            if domain and record.get("domain") != domain:
+                continue
+            results.append(self._to_domain_knowledge(record))
+
+        # Sort by confidence and return top_k
+        results.sort(key=lambda x: -x.confidence)
+        return results[:top_k]
+
+    def get_anti_patterns(
+        self,
+        project_id: str,
+        agent: Optional[str] = None,
+        embedding: Optional[List[float]] = None,
+        top_k: int = 5,
+    ) -> List[AntiPattern]:
+        """Get anti-patterns (no vector search)."""
+        data = self._read_json(self._files["anti_patterns"])
+
+        results = []
+        for record in data:
+            if record.get("project_id") != project_id:
+                continue
+            if agent and record.get("agent") != agent:
+                continue
+            results.append(self._to_anti_pattern(record))
+
+        # Sort by occurrence count and return top_k
+        results.sort(key=lambda x: -x.occurrence_count)
+        return results[:top_k]
+
+    # ==================== UPDATE OPERATIONS ====================
+
+    def update_heuristic(
+        self,
+        heuristic_id: str,
+        updates: Dict[str, Any],
+    ) -> bool:
+        """Update a heuristic's fields."""
+        data = self._read_json(self._files["heuristics"])
+
+        for i, record in enumerate(data):
+            if record.get("id") == heuristic_id:
+                data[i].update(updates)
+                self._write_json(self._files["heuristics"], data)
+                return True
+
+        return False
+
+    def increment_heuristic_occurrence(
+        self,
+        heuristic_id: str,
+        success: bool,
+    ) -> bool:
+        """Increment heuristic occurrence count."""
+        data = self._read_json(self._files["heuristics"])
+
+        for i, record in enumerate(data):
+            if record.get("id") == heuristic_id:
+                data[i]["occurrence_count"] = record.get("occurrence_count", 0) + 1
+                if success:
+                    data[i]["success_count"] = record.get("success_count", 0) + 1
+                data[i]["last_validated"] = datetime.now(timezone.utc).isoformat()
+                self._write_json(self._files["heuristics"], data)
+                return True
+
+        return False
+
+    # ==================== UPDATE CONFIDENCE OPERATIONS ====================
+
+    def update_heuristic_confidence(
+        self,
+        heuristic_id: str,
+        new_confidence: float,
+    ) -> bool:
+        """Update a heuristic's confidence score."""
+        data = self._read_json(self._files["heuristics"])
+
+        for i, record in enumerate(data):
+            if record.get("id") == heuristic_id:
+                data[i]["confidence"] = new_confidence
+                data[i]["last_validated"] = datetime.now(timezone.utc).isoformat()
+                self._write_json(self._files["heuristics"], data)
+                return True
+
+        return False
+
+    def update_knowledge_confidence(
+        self,
+        knowledge_id: str,
+        new_confidence: float,
+    ) -> bool:
+        """Update domain knowledge confidence score."""
+        data = self._read_json(self._files["domain_knowledge"])
+
+        for i, record in enumerate(data):
+            if record.get("id") == knowledge_id:
+                data[i]["confidence"] = new_confidence
+                data[i]["last_verified"] = datetime.now(timezone.utc).isoformat()
+                self._write_json(self._files["domain_knowledge"], data)
+                return True
+
+        return False
+
+    # ==================== DELETE OPERATIONS ====================
+
+    def delete_heuristic(self, heuristic_id: str) -> bool:
+        """Delete a single heuristic by ID."""
+        data = self._read_json(self._files["heuristics"])
+        original_count = len(data)
+
+        filtered = [r for r in data if r.get("id") != heuristic_id]
+        self._write_json(self._files["heuristics"], filtered)
+
+        deleted = original_count != len(filtered)
+        if deleted:
+            logger.debug(f"Deleted heuristic: {heuristic_id}")
+        return deleted
+
+    def delete_outcome(self, outcome_id: str) -> bool:
+        """Delete a single outcome by ID."""
+        data = self._read_json(self._files["outcomes"])
+        original_count = len(data)
+
+        filtered = [r for r in data if r.get("id") != outcome_id]
+        self._write_json(self._files["outcomes"], filtered)
+
+        deleted = original_count != len(filtered)
+        if deleted:
+            logger.debug(f"Deleted outcome: {outcome_id}")
+        return deleted
+
+    def delete_domain_knowledge(self, knowledge_id: str) -> bool:
+        """Delete a single domain knowledge entry by ID."""
+        data = self._read_json(self._files["domain_knowledge"])
+        original_count = len(data)
+
+        filtered = [r for r in data if r.get("id") != knowledge_id]
+        self._write_json(self._files["domain_knowledge"], filtered)
+
+        deleted = original_count != len(filtered)
+        if deleted:
+            logger.debug(f"Deleted domain knowledge: {knowledge_id}")
+        return deleted
+
+    def delete_anti_pattern(self, anti_pattern_id: str) -> bool:
+        """Delete a single anti-pattern by ID."""
+        data = self._read_json(self._files["anti_patterns"])
+        original_count = len(data)
+
+        filtered = [r for r in data if r.get("id") != anti_pattern_id]
+        self._write_json(self._files["anti_patterns"], filtered)
+
+        deleted = original_count != len(filtered)
+        if deleted:
+            logger.debug(f"Deleted anti-pattern: {anti_pattern_id}")
+        return deleted
+
+    def delete_outcomes_older_than(
+        self,
+        project_id: str,
+        older_than: datetime,
+        agent: Optional[str] = None,
+    ) -> int:
+        """Delete old outcomes."""
+        data = self._read_json(self._files["outcomes"])
+        original_count = len(data)
+
+        filtered = []
+        for record in data:
+            if record.get("project_id") != project_id:
+                filtered.append(record)
+                continue
+            if agent and record.get("agent") != agent:
+                filtered.append(record)
+                continue
+
+            timestamp = self._parse_datetime(record.get("timestamp"))
+            if timestamp and timestamp >= older_than:
+                filtered.append(record)
+
+        self._write_json(self._files["outcomes"], filtered)
+        deleted = original_count - len(filtered)
+        logger.info(f"Deleted {deleted} old outcomes")
+        return deleted
+
+    def delete_low_confidence_heuristics(
+        self,
+        project_id: str,
+        below_confidence: float,
+        agent: Optional[str] = None,
+    ) -> int:
+        """Delete low-confidence heuristics."""
+        data = self._read_json(self._files["heuristics"])
+        original_count = len(data)
+
+        filtered = []
+        for record in data:
+            if record.get("project_id") != project_id:
+                filtered.append(record)
+                continue
+            if agent and record.get("agent") != agent:
+                filtered.append(record)
+                continue
+
+            if record.get("confidence", 0) >= below_confidence:
+                filtered.append(record)
+
+        self._write_json(self._files["heuristics"], filtered)
+        deleted = original_count - len(filtered)
+        logger.info(f"Deleted {deleted} low-confidence heuristics")
+        return deleted
+
+    # ==================== STATS ====================
+
+    def get_stats(
+        self,
+        project_id: str,
+        agent: Optional[str] = None,
+    ) -> Dict[str, Any]:
+        """Get memory statistics."""
+        stats = {
+            "project_id": project_id,
+            "agent": agent,
+            "heuristics_count": 0,
+            "outcomes_count": 0,
+            "preferences_count": 0,
+            "domain_knowledge_count": 0,
+            "anti_patterns_count": 0,
+        }
+
+        for name, file_path in self._files.items():
+            data = self._read_json(file_path)
+            count = 0
+            for record in data:
+                if name == "preferences":
+                    # Preferences don't have project_id
+                    count += 1
+                elif record.get("project_id") == project_id:
+                    if agent is None or record.get("agent") == agent:
+                        count += 1
+            stats[f"{name}_count"] = count
+
+        stats["total_count"] = sum(stats[k] for k in stats if k.endswith("_count"))
+
+        return stats
+
+    # ==================== HELPERS ====================
+
+    def _read_json(self, file_path: Path) -> List[Dict]:
+        """Read JSON file."""
+        try:
+            with open(file_path, "r") as f:
+                return json.load(f)
+        except (json.JSONDecodeError, FileNotFoundError):
+            return []
+
+    def _write_json(self, file_path: Path, data: List[Dict]):
+        """Write JSON file."""
+        with open(file_path, "w") as f:
+            json.dump(data, f, indent=2, default=str)
+
+    def _to_dict(self, obj: Any) -> Dict:
+        """Convert dataclass to dict with datetime handling."""
+        if hasattr(obj, "__dataclass_fields__"):
+            result = {}
+            for field_name in obj.__dataclass_fields__:
+                value = getattr(obj, field_name)
+                if isinstance(value, datetime):
+                    result[field_name] = value.isoformat()
+                elif value is not None:
+                    result[field_name] = value
+            return result
+        return dict(obj)
+
+    def _parse_datetime(self, value: Any) -> Optional[datetime]:
+        """Parse datetime from string or return as-is."""
+        if value is None:
+            return None
+        if isinstance(value, datetime):
+            return value
+        try:
+            return datetime.fromisoformat(value.replace("Z", "+00:00"))
+        except (ValueError, AttributeError):
+            return None
+
+    def _to_heuristic(self, record: Dict) -> Heuristic:
+        """Convert dict to Heuristic."""
+        return Heuristic(
+            id=record["id"],
+            agent=record["agent"],
+            project_id=record["project_id"],
+            condition=record["condition"],
+            strategy=record["strategy"],
+            confidence=record.get("confidence", 0.0),
+            occurrence_count=record.get("occurrence_count", 0),
+            success_count=record.get("success_count", 0),
+            last_validated=self._parse_datetime(record.get("last_validated"))
+            or datetime.now(timezone.utc),
+            created_at=self._parse_datetime(record.get("created_at"))
+            or datetime.now(timezone.utc),
+            embedding=record.get("embedding"),
+            metadata=record.get("metadata", {}),
+        )
+
+    def _to_outcome(self, record: Dict) -> Outcome:
+        """Convert dict to Outcome."""
+        return Outcome(
+            id=record["id"],
+            agent=record["agent"],
+            project_id=record["project_id"],
+            task_type=record.get("task_type", "general"),
+            task_description=record["task_description"],
+            success=record.get("success", False),
+            strategy_used=record.get("strategy_used", ""),
+            duration_ms=record.get("duration_ms"),
+            error_message=record.get("error_message"),
+            user_feedback=record.get("user_feedback"),
+            timestamp=self._parse_datetime(record.get("timestamp"))
+            or datetime.now(timezone.utc),
+            embedding=record.get("embedding"),
+            metadata=record.get("metadata", {}),
+        )
+
+    def _to_user_preference(self, record: Dict) -> UserPreference:
+        """Convert dict to UserPreference."""
+        return UserPreference(
+            id=record["id"],
+            user_id=record["user_id"],
+            category=record.get("category", "general"),
+            preference=record["preference"],
+            source=record.get("source", "unknown"),
+            confidence=record.get("confidence", 1.0),
+            timestamp=self._parse_datetime(record.get("timestamp"))
+            or datetime.now(timezone.utc),
+            metadata=record.get("metadata", {}),
+        )
+
+    def _to_domain_knowledge(self, record: Dict) -> DomainKnowledge:
+        """Convert dict to DomainKnowledge."""
+        return DomainKnowledge(
+            id=record["id"],
+            agent=record["agent"],
+            project_id=record["project_id"],
+            domain=record.get("domain", "general"),
+            fact=record["fact"],
+            source=record.get("source", "unknown"),
+            confidence=record.get("confidence", 1.0),
+            last_verified=self._parse_datetime(record.get("last_verified"))
+            or datetime.now(timezone.utc),
+            embedding=record.get("embedding"),
+            metadata=record.get("metadata", {}),
+        )
+
+    def _to_anti_pattern(self, record: Dict) -> AntiPattern:
+        """Convert dict to AntiPattern."""
+        return AntiPattern(
+            id=record["id"],
+            agent=record["agent"],
+            project_id=record["project_id"],
+            pattern=record["pattern"],
+            why_bad=record.get("why_bad", ""),
+            better_alternative=record.get("better_alternative", ""),
+            occurrence_count=record.get("occurrence_count", 1),
+            last_seen=self._parse_datetime(record.get("last_seen"))
+            or datetime.now(timezone.utc),
+            created_at=self._parse_datetime(record.get("created_at"))
+            or datetime.now(timezone.utc),
+            embedding=record.get("embedding"),
+            metadata=record.get("metadata", {}),
+        )