refactor: standardize Redis key and index naming to dash convention

agent_memory_server/cli.py

@@ -21,6 +21,7 @@
     migrate_add_discrete_memory_extracted_2,
     migrate_add_memory_hashes_1,
     migrate_add_memory_type_3,
+    migrate_redis_key_naming_4,
 )
 from agent_memory_server.utils.redis import get_redis_conn
 
@@ -263,6 +264,81 @@ async def run_migration():
     asyncio.run(run_migration())
 
 
+@cli.command()
+@click.option(
+    "--batch-size",
+    default=50,
+    help="Number of keys to rename per pipeline batch",
+)
+@click.option(
+    "--dry-run",
+    is_flag=True,
+    help="Show what would change without executing",
+)
+def migrate_redis_naming(batch_size: int, dry_run: bool):
+    """
+    Migrate Redis key and index names from underscore to dash convention.
+
+    Renames keys:
+      memory_idx:* → memory-idx:*
+      working_memory:* → working-memory:*
+      auth_token:* → auth-token:*
+      auth_tokens:list → auth-tokens:list
+
+    Drops old indexes and rebuilds with new names.
+
+    Use --dry-run to see what would change without making modifications.
+    """
+    import asyncio
+
+    from agent_memory_server.memory_vector_db import RedisVLMemoryVectorDatabase
+    from agent_memory_server.memory_vector_db_factory import get_memory_vector_db
+    from agent_memory_server.working_memory_index import rebuild_working_memory_index
+
+    configure_logging()
+
+    async def run_migration():
+        redis = await get_redis_conn()
+
+        if dry_run:
+            click.echo("Dry run — no changes will be made.\n")
+
+        counts = await migrate_redis_key_naming_4(
+            redis=redis, batch_size=batch_size, dry_run=dry_run
+        )
+
+        click.echo("\nKey rename summary:")
+        click.echo(f"  memory_idx keys:     {counts['memory_idx']}")
+        click.echo(f"  working_memory keys: {counts['working_memory']}")
+        click.echo(f"  auth_token keys:     {counts['auth_token']}")
+        click.echo(f"  auth_tokens:list:    {counts['auth_tokens_list']}")
+        click.echo(f"  indexes dropped:     {counts['indexes_dropped']}")
+
+        if dry_run:
+            click.echo("\nNo changes were made (dry run).")
+            return
+
+        # Rebuild indexes with new names
+        click.echo("\nRebuilding indexes with new names...")
+
+        # Rebuild long-term memory index
+        db = await get_memory_vector_db()
+        if isinstance(db, RedisVLMemoryVectorDatabase):
+            index = db.index
+            await index.create(overwrite=True)
+            click.echo(f"  Rebuilt long-term memory index: {index.name}")
+
+        # Rebuild working memory index
+        await rebuild_working_memory_index(redis)
+        click.echo(
+            f"  Rebuilt working memory index: {settings.working_memory_index_name}"
+        )
+
+        click.echo("\nMigration completed successfully.")
+
+    asyncio.run(run_migration())
+
+
 @cli.command()
 @click.option("--port", default=settings.port, help="Port to run the server on")
 @click.option("--host", default="0.0.0.0", help="Host to run the server on")

agent_memory_server/config.py

@@ -374,7 +374,7 @@ class Settings(BaseSettings):
     )
 
     # RedisVL configuration (used by default Redis factory)
-    redisvl_index_name: str = "memory_records"
+    redisvl_index_name: str = "memory-records"
 
     # The server indexes messages in long-term memory by default. If this
     # setting is enabled, we also extract discrete memories from message text
@@ -400,13 +400,13 @@ class Settings(BaseSettings):
     # TODO: Adapt to memory database settings
     redisvl_distance_metric: str = "COSINE"
     redisvl_vector_dimensions: str = "1536"
-    redisvl_index_prefix: str = "memory_idx"
+    redisvl_index_prefix: str = "memory-idx"
     redisvl_indexing_algorithm: str = "HNSW"
 
     # Working Memory Index Settings
     # Used for listing sessions via Redis Search instead of sorted sets
-    working_memory_index_name: str = "working_memory_idx"
-    working_memory_index_prefix: str = "working_memory:"
+    working_memory_index_name: str = "working-memory-idx"
+    working_memory_index_prefix: str = "working-memory:"
 
     # Deduplication Settings (Store-Time)
     # Distance threshold for semantic similarity when deduplicating at store time

agent_memory_server/memory_vector_db.py

@@ -610,7 +610,7 @@ async def add_memories(self, memories: list[MemoryRecord]) -> list[str]:
                 memory_ids.append(memory.id)
 
             # Load into Redis via RedisVL -- use id_field so keys are
-            # auto-generated with the index prefix (e.g. "memory_idx:<id>").
+            # auto-generated with the index prefix (e.g. "memory-idx:<id>").
             # Do NOT pass explicit keys, as that bypasses the prefix.
             await self._index.load(data_list, id_field="id_")
             return memory_ids

agent_memory_server/migrations.py

@@ -133,3 +133,159 @@ async def migrate_add_memory_type_3(redis: Redis | None = None) -> None:
         migrated_count += 1
 
     logger.info(f"Migration completed. Added memory_type to {migrated_count} memories")
+
+
+async def migrate_redis_key_naming_4(
+    redis: Redis | None = None,
+    batch_size: int = 50,
+    dry_run: bool = False,
+) -> dict[str, int]:
+    """
+    Migration 4: Rename Redis keys and drop old indexes to use dash convention.
+
+    Renames:
+      - memory_idx:* → memory-idx:*
+      - working_memory:* → working-memory:*
+      - auth_token:* → auth-token:*
+      - auth_tokens:list → auth-tokens:list
+
+    Drops old indexes (without deleting data):
+      - memory_records
+      - working_memory_idx
+
+    Args:
+        redis: Optional Redis client
+        batch_size: Number of keys to rename per pipeline batch
+        dry_run: If True, only count keys without renaming
+
+    Returns:
+        Dict with counts per category
+    """
+    logger.info("Starting Redis key naming migration (underscore → dash)")
+    redis = redis or await get_redis_conn()
+
+    counts: dict[str, int] = {
+        "memory_idx": 0,
+        "working_memory": 0,
+        "auth_token": 0,
+        "auth_tokens_list": 0,
+        "indexes_dropped": 0,
+    }
+
+    # Migration status keys to skip (they are themselves being renamed)
+    migration_status_keys = {
+        b"working_memory:migration:complete",
+        b"working-memory:migration:complete",
+        b"working_memory:migration:remaining",
+        b"working-memory:migration:remaining",
+    }
+
+    async def _scan_and_rename(
+        pattern: str, old_prefix: str, new_prefix: str, category: str
+    ) -> int:
+        """Scan for keys matching pattern and rename old_prefix to new_prefix."""
+        renamed = 0
+        cursor = 0
+        while True:
+            cursor, keys = await redis.scan(cursor=cursor, match=pattern, count=1000)
+            if not keys:
+                if cursor == 0:
+                    break
+                continue
+
+            # Filter out migration status keys for working_memory category
+            if category == "working_memory":
+                keys = [k for k in keys if k not in migration_status_keys]
+
+            if not keys:
+                if cursor == 0:
+                    break
+                continue
+
+            if dry_run:
+                renamed += len(keys)
+            else:
+                # Batch rename using pipeline
+                for i in range(0, len(keys), batch_size):
+                    batch = keys[i : i + batch_size]
+                    pipe = redis.pipeline()
+                    for key in batch:
+                        key_str = key.decode("utf-8") if isinstance(key, bytes) else key
+                        new_key = key_str.replace(old_prefix, new_prefix, 1)
+                        pipe.rename(key_str, new_key)
+                    await pipe.execute()
+                    renamed += len(batch)
+
+            if cursor == 0:
+                break
+
+        return renamed
+
+    # 1. Rename memory_idx:* → memory-idx:*
+    counts["memory_idx"] = await _scan_and_rename(
+        "memory_idx:*", "memory_idx:", "memory-idx:", "memory_idx"
+    )
+    logger.info(
+        f"{'Would rename' if dry_run else 'Renamed'} {counts['memory_idx']} memory_idx keys"
+    )
+
+    # 2. Rename working_memory:* → working-memory:*
+    counts["working_memory"] = await _scan_and_rename(
+        "working_memory:*", "working_memory:", "working-memory:", "working_memory"
+    )
+    logger.info(
+        f"{'Would rename' if dry_run else 'Renamed'} {counts['working_memory']} working_memory keys"
+    )
+
+    # 3. Rename auth_token:* → auth-token:*
+    counts["auth_token"] = await _scan_and_rename(
+        "auth_token:*", "auth_token:", "auth-token:", "auth_token"
+    )
+    logger.info(
+        f"{'Would rename' if dry_run else 'Renamed'} {counts['auth_token']} auth_token keys"
+    )
+
+    # 4. Rename auth_tokens:list → auth-tokens:list (single key)
+    if not dry_run:
+        exists = await redis.exists("auth_tokens:list")
+        if exists:
+            await redis.rename("auth_tokens:list", "auth-tokens:list")
+            counts["auth_tokens_list"] = 1
+            logger.info("Renamed auth_tokens:list → auth-tokens:list")
+    else:
+        exists = await redis.exists("auth_tokens:list")
+        if exists:
+            counts["auth_tokens_list"] = 1
+            logger.info("Would rename auth_tokens:list → auth-tokens:list")
+
+    # 5. Drop old indexes (FT.DROPINDEX without DD flag preserves data)
+    if not dry_run:
+        for old_index_name in ("memory_records", "working_memory_idx"):
+            try:
+                await redis.execute_command("FT.DROPINDEX", old_index_name)
+                counts["indexes_dropped"] += 1
+                logger.info(f"Dropped old index '{old_index_name}'")
+            except Exception as e:
+                # Index may not exist, which is fine
+                if "Unknown index name" in str(e) or "Unknown Index name" in str(e):
+                    logger.info(
+                        f"Old index '{old_index_name}' does not exist, skipping"
+                    )
+                else:
+                    logger.warning(f"Failed to drop index '{old_index_name}': {e}")
+    else:
+        for old_index_name in ("memory_records", "working_memory_idx"):
+            try:
+                await redis.execute_command("FT.INFO", old_index_name)
+                counts["indexes_dropped"] += 1
+                logger.info(f"Would drop old index '{old_index_name}'")
+            except Exception:
+                logger.info(f"Old index '{old_index_name}' does not exist, skipping")
+
+    total = sum(counts.values())
+    logger.info(
+        f"Redis key naming migration {'(dry run) ' if dry_run else ''}complete. "
+        f"Total: {total} operations ({counts})"
+    )
+
+    return counts

agent_memory_server/utils/keys.py

@@ -78,7 +78,7 @@ def working_memory_key(
     ) -> str:
         """Get the working memory key for a session."""
         # Build key components, filtering out None values
-        key_parts = ["working_memory"]
+        key_parts = ["working-memory"]
 
         if namespace:
             key_parts.append(namespace)
@@ -98,9 +98,9 @@ def search_index_name() -> str:
     @staticmethod
     def auth_token_key(token_hash: str) -> str:
         """Get the auth token key for a hashed token."""
-        return f"auth_token:{token_hash}"
+        return f"auth-token:{token_hash}"
 
     @staticmethod
     def auth_tokens_list_key() -> str:
         """Get the key for the list of all auth tokens."""
-        return "auth_tokens:list"
+        return "auth-tokens:list"

agent_memory_server/working_memory.py

@@ -23,8 +23,8 @@
 logger = logging.getLogger(__name__)
 
 # Redis keys for migration status (shared across workers, persists across restarts)
-MIGRATION_STATUS_KEY = "working_memory:migration:complete"
-MIGRATION_REMAINING_KEY = "working_memory:migration:remaining"
+MIGRATION_STATUS_KEY = "working-memory:migration:complete"
+MIGRATION_REMAINING_KEY = "working-memory:migration:remaining"
 
 
 async def check_and_set_migration_status(redis_client: Redis | None = None) -> bool:
@@ -62,7 +62,7 @@ async def check_and_set_migration_status(redis_client: Redis | None = None) -> b
             )
             return True
 
-    # Scan for working_memory:* keys of type STRING only
+    # Scan for working-memory:* keys of type STRING only
     # This is much faster than scanning all keys and calling TYPE on each
     cursor = 0
     string_keys_found = 0
@@ -71,7 +71,7 @@ async def check_and_set_migration_status(redis_client: Redis | None = None) -> b
         while True:
             # Use _type="string" to only get string keys directly
             cursor, keys = await redis_client.scan(
-                cursor=cursor, match="working_memory:*", count=1000, _type="string"
+                cursor=cursor, match="working-memory:*", count=1000, _type="string"
             )
 
             if keys:

agent_memory_server/working_memory_index.py

@@ -85,7 +85,7 @@ async def ensure_working_memory_index(redis_client: Redis) -> bool:
     """
     Ensure the working memory search index exists.
 
-    Creates a Redis Search index on JSON documents with prefix 'working_memory:'
+    Creates a Redis Search index on JSON documents with prefix 'working-memory:'
     if it doesn't already exist. The index enables efficient session listing
     with filtering by namespace and user_id.
 

docs/configuration.md

@@ -101,10 +101,10 @@ INDEX_ALL_MESSAGES_IN_LONG_TERM_MEMORY=false  # Index every message (default: fa
 MEMORY_VECTOR_DB_FACTORY=agent_memory_server.memory_vector_db_factory.create_redis_memory_vector_db
 
 # RedisVL Settings (used by default Redis factory)
-REDISVL_INDEX_NAME=memory_records        # Index name (default: memory_records)
+REDISVL_INDEX_NAME=memory-records        # Index name (default: memory-records)
 REDISVL_DISTANCE_METRIC=COSINE           # Distance metric (default: COSINE)
 REDISVL_VECTOR_DIMENSIONS=1536           # Vector dimensions (default: 1536)
-REDISVL_INDEX_PREFIX=memory_idx          # Index prefix (default: memory_idx)
+REDISVL_INDEX_PREFIX=memory-idx          # Index prefix (default: memory-idx)
 REDISVL_INDEXING_ALGORITHM=HNSW          # Indexing algorithm (default: HNSW)
 ```