Memory Service¶

This guide shows how to wire RedisLongTermMemoryService into a Google ADK agent for persistent long-term memory backed by the Agent Memory Server.

For the concepts behind long-term memory, see Sessions + Memory Services.

Prerequisites¶

Agent Memory Server running on localhost:8000 (see Memory server setup).
adk-redis with the memory extra: pip install "adk-redis[memory]".

Basic usage¶

from google.adk import Agent
from google.adk.runners import Runner
from adk_redis import (
    RedisLongTermMemoryService,
    RedisLongTermMemoryServiceConfig,
    RedisWorkingMemorySessionService,
    RedisWorkingMemorySessionServiceConfig,
)

session_service = RedisWorkingMemorySessionService(
    config=RedisWorkingMemorySessionServiceConfig(
        api_base_url="http://localhost:8000",
        default_namespace="my_app",
    )
)

memory_service = RedisLongTermMemoryService(
    config=RedisLongTermMemoryServiceConfig(
        api_base_url="http://localhost:8000",
        default_namespace="my_app",
        recency_boost=True,
    )
)

agent = Agent(
    model="gemini-2.0-flash",
    name="my_agent",
    instruction="You are a helpful assistant with memory.",
)

runner = Runner(
    agent=agent,
    app_name="my_app",
    session_service=session_service,
    memory_service=memory_service,
)

How memories flow¶

The agent converses with the user in a session.
The Agent Memory Server extracts key facts in the background.
Facts are stored as long-term memories with vector embeddings.
On future sessions, the runner calls search_memory() and injects relevant memories into the prompt automatically.

Configuration options¶

Option	Default	Description
`api_base_url`	`http://localhost:8000`	Agent Memory Server URL
`default_namespace`	`None`	Namespace for memory isolation
`search_top_k`	`10`	Max memories per search
`distance_threshold`	`None`	Max distance for results (0.0-1.0)
`recency_boost`	`True`	Enable recency-aware ranking
`semantic_weight`	`0.8`	Weight for semantic similarity
`recency_weight`	`0.2`	Weight for recency score
`extraction_strategy`	`discrete`	`discrete`, `summary`, `preferences`, `custom`
`timeout`	`30.0`	HTTP request timeout