claude-configs/skills/memorygraph/client.py

#!/usr/bin/env python3
"""
MemoryGraph SQLite Client

Direct SQLite interface for MemoryGraph operations, bypassing MCP protocol.
Provides CLI and Python API for memory storage, retrieval, and relationship management.

Usage:
    # CLI
    python client.py store --type solution --title "Fixed X" --content "Details..." --tags "python,fix"
    python client.py recall "timeout error"
    python client.py get <memory_id>
    python client.py relate <from_id> <to_id> SOLVES

    # Python
    from client import MemoryGraphClient
    client = MemoryGraphClient()
    memory_id = client.store_memory(type="solution", title="...", content="...")
"""

import argparse
import json
import os
import re
import sqlite3
import sys
import uuid
from datetime import datetime, timezone
from pathlib import Path
from typing import Any, Dict, List, Optional, Union


# Database location
DB_PATH = Path.home() / ".memorygraph" / "memory.db"

# Valid memory types
MEMORY_TYPES = [
    "solution", "problem", "error", "fix", "code_pattern", "decision",
    "configuration", "workflow", "task", "project", "technology",
    "command", "file_context", "general"
]

# Valid relationship types
RELATIONSHIP_TYPES = [
    # Causal
    "CAUSES", "TRIGGERS", "LEADS_TO", "PREVENTS", "BREAKS",
    # Solution
    "SOLVES", "ADDRESSES", "ALTERNATIVE_TO", "IMPROVES", "REPLACES",
    # Context
    "OCCURS_IN", "APPLIES_TO", "WORKS_WITH", "REQUIRES", "USED_IN",
    # Learning
    "BUILDS_ON", "CONTRADICTS", "CONFIRMS", "GENERALIZES", "SPECIALIZES",
    # Similarity
    "SIMILAR_TO", "VARIANT_OF", "RELATED_TO", "ANALOGY_TO", "OPPOSITE_OF",
    # Workflow
    "FOLLOWS", "DEPENDS_ON", "ENABLES", "BLOCKS", "PARALLEL_TO",
    # Quality
    "EFFECTIVE_FOR", "INEFFECTIVE_FOR", "PREFERRED_OVER", "DEPRECATED_BY", "VALIDATED_BY"
]


class MemoryGraphClient:
    """Client for direct SQLite access to MemoryGraph database."""

    def __init__(self, db_path: Optional[Path] = None):
        self.db_path = db_path or DB_PATH
        if not self.db_path.exists():
            raise FileNotFoundError(f"Database not found: {self.db_path}")

    def _connect(self) -> sqlite3.Connection:
        """Create database connection with row factory."""
        conn = sqlite3.connect(str(self.db_path))
        conn.row_factory = sqlite3.Row
        return conn

    def _now_iso(self) -> str:
        """Get current timestamp in ISO format."""
        return datetime.now(timezone.utc).isoformat()

    # =========================================================================
    # CORE OPERATIONS (High Priority)
    # =========================================================================

    def store_memory(
        self,
        type: str,
        title: str,
        content: str,
        tags: Optional[List[str]] = None,
        importance: float = 0.5,
        summary: Optional[str] = None,
        context: Optional[Dict[str, Any]] = None
    ) -> str:
        """
        Store a new memory.

        Args:
            type: Memory type (solution, problem, error, fix, etc.)
            title: Short descriptive title
            content: Detailed content
            tags: List of tags for categorization
            importance: Importance score 0.0-1.0 (default 0.5)
            summary: Optional brief summary
            context: Optional context dict (project_path, files, etc.)

        Returns:
            memory_id: UUID of the created memory
        """
        if type not in MEMORY_TYPES:
            raise ValueError(f"Invalid memory type: {type}. Valid types: {MEMORY_TYPES}")

        memory_id = str(uuid.uuid4())
        now = self._now_iso()

        # Normalize tags to lowercase
        tags = [t.lower().strip() for t in (tags or [])]

        properties = {
            "id": memory_id,
            "type": type,
            "title": title,
            "content": content,
            "summary": summary,
            "tags": tags,
            "importance": max(0.0, min(1.0, importance)),
            "confidence": 0.8,
            "usage_count": 0,
            "created_at": now,
            "updated_at": now
        }

        if context:
            properties["context"] = context

        with self._connect() as conn:
            # Insert node
            conn.execute(
                "INSERT INTO nodes (id, label, properties, created_at, updated_at) VALUES (?, ?, ?, ?, ?)",
                (memory_id, "Memory", json.dumps(properties), now, now)
            )

            # Note: FTS is an external content table managed by triggers in the MCP schema.
            # We use JSON-based search instead, so no FTS update needed.

            conn.commit()

        return memory_id

    def recall_memories(
        self,
        query: str,
        memory_types: Optional[List[str]] = None,
        limit: int = 20,
        project_path: Optional[str] = None
    ) -> List[Dict[str, Any]]:
        """
        Natural language search for memories with fuzzy matching.

        Args:
            query: Natural language search query
            memory_types: Optional filter by memory types
            limit: Maximum results (default 20)
            project_path: Optional filter by project path

        Returns:
            List of matching memories with relevance info
        """
        query_lower = query.lower().strip()
        terms = query_lower.split()

        with self._connect() as conn:
            # Use JSON-based search (more reliable than FTS for existing data)
            # Search in title, content, summary, and tags
            rows = conn.execute(
                """
                SELECT id, properties, created_at
                FROM nodes
                WHERE label = 'Memory'
                ORDER BY created_at DESC
                """
            ).fetchall()

            results = []
            for row in rows:
                props = json.loads(row["properties"])

                # Filter by memory type if specified
                if memory_types and props.get("type") not in memory_types:
                    continue

                # Filter by project path if specified
                if project_path:
                    ctx = props.get("context", {})
                    if isinstance(ctx, dict) and ctx.get("project_path") != project_path:
                        continue

                # Score based on term matches in title, content, summary, tags
                title = (props.get("title") or "").lower()
                content = (props.get("content") or "").lower()
                summary = (props.get("summary") or "").lower()
                tags = " ".join(props.get("tags") or []).lower()
                searchable = f"{title} {content} {summary} {tags}"

                # Check if any term matches
                matches = sum(1 for term in terms if term in searchable)
                if matches == 0:
                    continue

                # Get relationships for context
                rels = self._get_relationships_for_memory(conn, row["id"])

                # Calculate relevance score
                title_matches = sum(1 for term in terms if term in title) * 3
                tag_matches = sum(1 for term in terms if term in tags) * 2
                content_matches = matches
                score = title_matches + tag_matches + content_matches

                results.append({
                    "id": props["id"],
                    "type": props.get("type"),
                    "title": props.get("title"),
                    "content": props.get("content"),
                    "summary": props.get("summary"),
                    "tags": props.get("tags", []),
                    "importance": props.get("importance"),
                    "created_at": props.get("created_at"),
                    "relationships": rels,
                    "match_query": query,
                    "_score": score
                })

            # Sort by score (higher = better match)
            results.sort(key=lambda x: x.pop("_score", 0), reverse=True)
            return results[:limit]

    def get_memory(
        self,
        memory_id: str,
        include_relationships: bool = True
    ) -> Optional[Dict[str, Any]]:
        """
        Retrieve a specific memory by ID.

        Args:
            memory_id: UUID of the memory
            include_relationships: Whether to include related memories

        Returns:
            Memory dict or None if not found
        """
        with self._connect() as conn:
            row = conn.execute(
                "SELECT id, properties, created_at, updated_at FROM nodes WHERE id = ? AND label = 'Memory'",
                (memory_id,)
            ).fetchone()

            if not row:
                return None

            props = json.loads(row["properties"])

            result = {
                "id": props["id"],
                "type": props.get("type"),
                "title": props.get("title"),
                "content": props.get("content"),
                "summary": props.get("summary"),
                "tags": props.get("tags", []),
                "importance": props.get("importance"),
                "confidence": props.get("confidence"),
                "usage_count": props.get("usage_count", 0),
                "created_at": props.get("created_at"),
                "updated_at": props.get("updated_at"),
                "context": props.get("context")
            }

            if include_relationships:
                result["relationships"] = self._get_relationships_for_memory(conn, memory_id)

            # Increment usage count
            props["usage_count"] = props.get("usage_count", 0) + 1
            conn.execute(
                "UPDATE nodes SET properties = ? WHERE id = ?",
                (json.dumps(props), memory_id)
            )
            conn.commit()

            return result

    def create_relationship(
        self,
        from_memory_id: str,
        to_memory_id: str,
        relationship_type: str,
        strength: float = 0.5,
        confidence: float = 0.8,
        context: Optional[str] = None
    ) -> str:
        """
        Create a relationship between two memories.

        Args:
            from_memory_id: Source memory UUID
            to_memory_id: Target memory UUID
            relationship_type: Type of relationship (SOLVES, CAUSES, etc.)
            strength: Relationship strength 0.0-1.0 (default 0.5)
            confidence: Confidence score 0.0-1.0 (default 0.8)
            context: Optional context description

        Returns:
            relationship_id: UUID of the created relationship
        """
        if relationship_type not in RELATIONSHIP_TYPES:
            raise ValueError(f"Invalid relationship type: {relationship_type}")

        rel_id = str(uuid.uuid4())
        now = self._now_iso()

        # Build context structure
        context_obj = {
            "text": context or "",
            "scope": None,
            "components": [],
            "conditions": [],
            "evidence": [],
            "temporal": None,
            "exceptions": []
        }

        properties = {
            "id": rel_id,
            "strength": max(0.0, min(1.0, strength)),
            "confidence": max(0.0, min(1.0, confidence)),
            "context": json.dumps(context_obj),
            "evidence_count": 1 if context else 0,
            "success_rate": None,
            "created_at": now,
            "last_validated": now,
            "validation_count": 0,
            "counter_evidence_count": 0
        }

        with self._connect() as conn:
            # Verify both memories exist
            for mid in [from_memory_id, to_memory_id]:
                exists = conn.execute(
                    "SELECT 1 FROM nodes WHERE id = ? AND label = 'Memory'", (mid,)
                ).fetchone()
                if not exists:
                    raise ValueError(f"Memory not found: {mid}")

            conn.execute(
                "INSERT INTO relationships (id, from_id, to_id, rel_type, properties, created_at) VALUES (?, ?, ?, ?, ?, ?)",
                (rel_id, from_memory_id, to_memory_id, relationship_type, json.dumps(properties), now)
            )
            conn.commit()

        return rel_id

    # =========================================================================
    # MEDIUM PRIORITY OPERATIONS
    # =========================================================================

    def search_memories(
        self,
        query: Optional[str] = None,
        memory_types: Optional[List[str]] = None,
        tags: Optional[List[str]] = None,
        min_importance: Optional[float] = None,
        limit: int = 50,
        offset: int = 0
    ) -> List[Dict[str, Any]]:
        """
        Advanced search with fine-grained control.

        Args:
            query: Text search query (optional)
            memory_types: Filter by memory types
            tags: Filter by tags (matches any)
            min_importance: Minimum importance score
            limit: Maximum results
            offset: Pagination offset

        Returns:
            List of matching memories
        """
        query_terms = query.lower().strip().split() if query else []

        with self._connect() as conn:
            rows = conn.execute(
                "SELECT id, properties FROM nodes WHERE label = 'Memory' ORDER BY created_at DESC"
            ).fetchall()

            results = []
            for row in rows:
                props = json.loads(row["properties"])

                # Apply type filter
                if memory_types and props.get("type") not in memory_types:
                    continue

                # Apply importance filter
                if min_importance and props.get("importance", 0) < min_importance:
                    continue

                # Apply tag filter
                if tags:
                    mem_tags = set(props.get("tags", []))
                    if not mem_tags.intersection(set(t.lower() for t in tags)):
                        continue

                # Apply query filter
                if query_terms:
                    title = (props.get("title") or "").lower()
                    content = (props.get("content") or "").lower()
                    summary = (props.get("summary") or "").lower()
                    mem_tags_str = " ".join(props.get("tags") or []).lower()
                    searchable = f"{title} {content} {summary} {mem_tags_str}"
                    if not any(term in searchable for term in query_terms):
                        continue

                content_preview = props.get("content") or ""
                if len(content_preview) > 200:
                    content_preview = content_preview[:200] + "..."

                results.append({
                    "id": props["id"],
                    "type": props.get("type"),
                    "title": props.get("title"),
                    "content": content_preview,
                    "tags": props.get("tags", []),
                    "importance": props.get("importance"),
                    "created_at": props.get("created_at")
                })

            # Apply pagination
            return results[offset:offset + limit]

    def update_memory(
        self,
        memory_id: str,
        title: Optional[str] = None,
        content: Optional[str] = None,
        summary: Optional[str] = None,
        tags: Optional[List[str]] = None,
        importance: Optional[float] = None
    ) -> bool:
        """
        Update an existing memory.

        Args:
            memory_id: UUID of memory to update
            title, content, summary, tags, importance: Fields to update (None = no change)

        Returns:
            True if updated, False if not found
        """
        with self._connect() as conn:
            row = conn.execute(
                "SELECT properties FROM nodes WHERE id = ? AND label = 'Memory'",
                (memory_id,)
            ).fetchone()

            if not row:
                return False

            props = json.loads(row["properties"])
            now = self._now_iso()

            # Update fields
            if title is not None:
                props["title"] = title
            if content is not None:
                props["content"] = content
            if summary is not None:
                props["summary"] = summary
            if tags is not None:
                props["tags"] = [t.lower().strip() for t in tags]
            if importance is not None:
                props["importance"] = max(0.0, min(1.0, importance))

            props["updated_at"] = now

            conn.execute(
                "UPDATE nodes SET properties = ?, updated_at = ? WHERE id = ?",
                (json.dumps(props), now, memory_id)
            )

            # Note: FTS is an external content table - we use JSON-based search instead

            conn.commit()
            return True

    def delete_memory(self, memory_id: str) -> bool:
        """
        Delete a memory and all its relationships.

        Args:
            memory_id: UUID of memory to delete

        Returns:
            True if deleted, False if not found
        """
        with self._connect() as conn:
            # Check exists
            exists = conn.execute(
                "SELECT 1 FROM nodes WHERE id = ? AND label = 'Memory'", (memory_id,)
            ).fetchone()

            if not exists:
                return False

            # Delete relationships (CASCADE should handle but be explicit)
            conn.execute(
                "DELETE FROM relationships WHERE from_id = ? OR to_id = ?",
                (memory_id, memory_id)
            )

            # Delete node (FTS is external content table - no direct deletion needed)
            conn.execute("DELETE FROM nodes WHERE id = ?", (memory_id,))

            conn.commit()
            return True

    def get_related_memories(
        self,
        memory_id: str,
        relationship_types: Optional[List[str]] = None,
        max_depth: int = 1
    ) -> List[Dict[str, Any]]:
        """
        Find memories connected to a specific memory.

        Args:
            memory_id: UUID of the memory
            relationship_types: Filter by relationship types
            max_depth: Traversal depth (1-5)

        Returns:
            List of related memories with relationship info
        """
        max_depth = max(1, min(5, max_depth))

        with self._connect() as conn:
            visited = set()
            results = []

            def traverse(mid: str, depth: int):
                if depth > max_depth or mid in visited:
                    return
                visited.add(mid)

                # Get outgoing relationships
                rows = conn.execute(
                    """
                    SELECT r.rel_type, r.properties as rel_props, n.id, n.properties
                    FROM relationships r
                    JOIN nodes n ON r.to_id = n.id
                    WHERE r.from_id = ? AND n.label = 'Memory'
                    """,
                    (mid,)
                ).fetchall()

                for row in rows:
                    if relationship_types and row["rel_type"] not in relationship_types:
                        continue
                    if row["id"] not in visited:
                        props = json.loads(row["properties"])
                        rel_props = json.loads(row["rel_props"])
                        results.append({
                            "id": props["id"],
                            "type": props.get("type"),
                            "title": props.get("title"),
                            "relationship": row["rel_type"],
                            "direction": "outgoing",
                            "strength": rel_props.get("strength"),
                            "depth": depth
                        })
                        traverse(row["id"], depth + 1)

                # Get incoming relationships
                rows = conn.execute(
                    """
                    SELECT r.rel_type, r.properties as rel_props, n.id, n.properties
                    FROM relationships r
                    JOIN nodes n ON r.from_id = n.id
                    WHERE r.to_id = ? AND n.label = 'Memory'
                    """,
                    (mid,)
                ).fetchall()

                for row in rows:
                    if relationship_types and row["rel_type"] not in relationship_types:
                        continue
                    if row["id"] not in visited:
                        props = json.loads(row["properties"])
                        rel_props = json.loads(row["rel_props"])
                        results.append({
                            "id": props["id"],
                            "type": props.get("type"),
                            "title": props.get("title"),
                            "relationship": row["rel_type"],
                            "direction": "incoming",
                            "strength": rel_props.get("strength"),
                            "depth": depth
                        })
                        traverse(row["id"], depth + 1)

            traverse(memory_id, 1)
            return results

    # =========================================================================
    # LOW PRIORITY OPERATIONS
    # =========================================================================

    def get_memory_statistics(self) -> Dict[str, Any]:
        """Get database statistics."""
        with self._connect() as conn:
            # Total memories
            total = conn.execute(
                "SELECT COUNT(*) as count FROM nodes WHERE label = 'Memory'"
            ).fetchone()["count"]

            # Count by type
            rows = conn.execute(
                """
                SELECT json_extract(properties, '$.type') as type, COUNT(*) as count
                FROM nodes WHERE label = 'Memory'
                GROUP BY json_extract(properties, '$.type')
                """
            ).fetchall()
            by_type = {row["type"]: row["count"] for row in rows}

            # Relationship count
            rel_count = conn.execute(
                "SELECT COUNT(*) as count FROM relationships"
            ).fetchone()["count"]

            # DB size
            db_size = self.db_path.stat().st_size

            return {
                "total_memories": total,
                "by_type": by_type,
                "total_relationships": rel_count,
                "database_size_bytes": db_size,
                "database_path": str(self.db_path)
            }

    def get_recent_activity(
        self,
        days: int = 7,
        project: Optional[str] = None
    ) -> Dict[str, Any]:
        """Get recent memory activity summary."""
        with self._connect() as conn:
            cutoff = datetime.now(timezone.utc).isoformat()[:10]  # Just date part for simplicity

            rows = conn.execute(
                """
                SELECT id, properties FROM nodes
                WHERE label = 'Memory'
                ORDER BY created_at DESC
                LIMIT 50
                """
            ).fetchall()

            recent = []
            by_type = {}
            for row in rows:
                props = json.loads(row["properties"])
                mem_type = props.get("type", "unknown")
                by_type[mem_type] = by_type.get(mem_type, 0) + 1
                recent.append({
                    "id": props["id"],
                    "type": mem_type,
                    "title": props.get("title"),
                    "created_at": props.get("created_at")
                })

            return {
                "period_days": days,
                "summary_by_type": by_type,
                "recent_memories": recent[:20],
                "total_in_period": len(recent)
            }

    # =========================================================================
    # HELPER METHODS
    # =========================================================================

    def _get_relationships_for_memory(self, conn: sqlite3.Connection, memory_id: str) -> Dict[str, List[str]]:
        """Get relationships grouped by type for a memory."""
        rels = {}

        # Outgoing
        rows = conn.execute(
            "SELECT rel_type, to_id FROM relationships WHERE from_id = ?",
            (memory_id,)
        ).fetchall()
        for row in rows:
            key = f"{row['rel_type']}_to"
            if key not in rels:
                rels[key] = []
            rels[key].append(row["to_id"])

        # Incoming
        rows = conn.execute(
            "SELECT rel_type, from_id FROM relationships WHERE to_id = ?",
            (memory_id,)
        ).fetchall()
        for row in rows:
            key = f"{row['rel_type']}_from"
            if key not in rels:
                rels[key] = []
            rels[key].append(row["from_id"])

        return rels


# =============================================================================
# CLI INTERFACE
# =============================================================================

def main():
    parser = argparse.ArgumentParser(
        description="MemoryGraph CLI - Direct SQLite interface for memory operations",
        formatter_class=argparse.RawDescriptionHelpFormatter
    )
    subparsers = parser.add_subparsers(dest="command", help="Commands")

    # store
    store_parser = subparsers.add_parser("store", help="Store a new memory")
    store_parser.add_argument("--type", "-t", required=True, choices=MEMORY_TYPES, help="Memory type")
    store_parser.add_argument("--title", required=True, help="Memory title")
    store_parser.add_argument("--content", "-c", required=True, help="Memory content")
    store_parser.add_argument("--tags", help="Comma-separated tags")
    store_parser.add_argument("--importance", "-i", type=float, default=0.5, help="Importance 0.0-1.0")
    store_parser.add_argument("--summary", "-s", help="Brief summary")

    # recall
    recall_parser = subparsers.add_parser("recall", help="Search memories with natural language")
    recall_parser.add_argument("query", help="Search query")
    recall_parser.add_argument("--types", help="Comma-separated memory types to filter")
    recall_parser.add_argument("--limit", "-n", type=int, default=10, help="Max results")

    # get
    get_parser = subparsers.add_parser("get", help="Get a specific memory by ID")
    get_parser.add_argument("memory_id", help="Memory UUID")
    get_parser.add_argument("--no-rels", action="store_true", help="Exclude relationships")

    # relate
    relate_parser = subparsers.add_parser("relate", help="Create relationship between memories")
    relate_parser.add_argument("from_id", help="Source memory UUID")
    relate_parser.add_argument("to_id", help="Target memory UUID")
    relate_parser.add_argument("rel_type", choices=RELATIONSHIP_TYPES, help="Relationship type")
    relate_parser.add_argument("--context", help="Context description")
    relate_parser.add_argument("--strength", type=float, default=0.5, help="Strength 0.0-1.0")

    # search
    search_parser = subparsers.add_parser("search", help="Advanced search with filters")
    search_parser.add_argument("--query", "-q", help="Text query")
    search_parser.add_argument("--types", help="Comma-separated memory types")
    search_parser.add_argument("--tags", help="Comma-separated tags")
    search_parser.add_argument("--min-importance", type=float, help="Minimum importance")
    search_parser.add_argument("--limit", "-n", type=int, default=20, help="Max results")

    # update
    update_parser = subparsers.add_parser("update", help="Update a memory")
    update_parser.add_argument("memory_id", help="Memory UUID")
    update_parser.add_argument("--title", help="New title")
    update_parser.add_argument("--content", help="New content")
    update_parser.add_argument("--tags", help="New tags (comma-separated)")
    update_parser.add_argument("--importance", type=float, help="New importance")

    # delete
    delete_parser = subparsers.add_parser("delete", help="Delete a memory")
    delete_parser.add_argument("memory_id", help="Memory UUID")
    delete_parser.add_argument("--force", "-f", action="store_true", help="Skip confirmation")

    # related
    related_parser = subparsers.add_parser("related", help="Get related memories")
    related_parser.add_argument("memory_id", help="Memory UUID")
    related_parser.add_argument("--types", help="Comma-separated relationship types")
    related_parser.add_argument("--depth", type=int, default=1, help="Traversal depth 1-5")

    # stats
    subparsers.add_parser("stats", help="Show database statistics")

    # recent
    recent_parser = subparsers.add_parser("recent", help="Show recent activity")
    recent_parser.add_argument("--days", type=int, default=7, help="Days to look back")

    args = parser.parse_args()

    if not args.command:
        parser.print_help()
        sys.exit(1)

    try:
        client = MemoryGraphClient()
    except FileNotFoundError as e:
        print(json.dumps({"error": str(e)}))
        sys.exit(1)

    result = None

    if args.command == "store":
        tags = [t.strip() for t in args.tags.split(",")] if args.tags else None
        memory_id = client.store_memory(
            type=args.type,
            title=args.title,
            content=args.content,
            tags=tags,
            importance=args.importance,
            summary=args.summary
        )
        result = {"success": True, "memory_id": memory_id}

    elif args.command == "recall":
        types = [t.strip() for t in args.types.split(",")] if args.types else None
        result = client.recall_memories(args.query, memory_types=types, limit=args.limit)

    elif args.command == "get":
        result = client.get_memory(args.memory_id, include_relationships=not args.no_rels)
        if not result:
            result = {"error": "Memory not found"}

    elif args.command == "relate":
        rel_id = client.create_relationship(
            args.from_id, args.to_id, args.rel_type,
            strength=args.strength, context=args.context
        )
        result = {"success": True, "relationship_id": rel_id}

    elif args.command == "search":
        types = [t.strip() for t in args.types.split(",")] if args.types else None
        tags = [t.strip() for t in args.tags.split(",")] if args.tags else None
        result = client.search_memories(
            query=args.query,
            memory_types=types,
            tags=tags,
            min_importance=args.min_importance,
            limit=args.limit
        )

    elif args.command == "update":
        tags = [t.strip() for t in args.tags.split(",")] if args.tags else None
        success = client.update_memory(
            args.memory_id,
            title=args.title,
            content=args.content,
            tags=tags,
            importance=args.importance
        )
        result = {"success": success}

    elif args.command == "delete":
        if not args.force:
            # Get memory first to show what's being deleted
            mem = client.get_memory(args.memory_id, include_relationships=False)
            if mem:
                print(f"Deleting: {mem.get('title')}", file=sys.stderr)
        success = client.delete_memory(args.memory_id)
        result = {"success": success}

    elif args.command == "related":
        types = [t.strip() for t in args.types.split(",")] if args.types else None
        result = client.get_related_memories(args.memory_id, relationship_types=types, max_depth=args.depth)

    elif args.command == "stats":
        result = client.get_memory_statistics()

    elif args.command == "recent":
        result = client.get_recent_activity(days=args.days)

    print(json.dumps(result, indent=2, default=str))


if __name__ == "__main__":
    main()