Redis Agent Memory Server provides a fast, flexible memory layer for AI agents and applications, leveraging Redis for persistence and retrieval. It targets developers building stateful AI systems, offering benefits like persistent conversational context, user preference learning, and knowledge accumulation across agent interactions.

How It Works

The server employs a two-tier memory architecture: session-scoped working memory and persistent long-term memory. It exposes both a REST API and a Model Context Protocol (MCP) server for integration. Core functionalities include configurable memory extraction strategies (e.g., discrete, summary), and versatile search capabilities encompassing semantic (vector-based similarity), keyword, and hybrid approaches, all filterable by metadata. A pluggable backend system supports various vector databases, and multi-provider LLM support via LiteLLM enables features like automatic topic extraction and entity recognition.

Quick Start & Requirements

Installation is available via pre-built Docker images (redislabs/agent-memory-server or ghcr.io/redis/agent-memory-server) or from source using pip install uv and uv sync --all-extras. A typical development setup involves running docker compose up api redis or uv run agent-memory api --task-backend=asyncio. Production deployments require separate containers for the API server and background workers. Key prerequisites include a running Redis instance and API keys for desired LLM providers (e.g., OPENAI_API_KEY). Official documentation is available at Documentation.

Highlighted Details

• Dual Interface: Supports both REST API and Model Context Protocol (MCP) server.
• Two-Tier Memory: Manages both session-scoped working memory and persistent long-term memory.
• Advanced Search: Offers semantic, keyword, and hybrid search with metadata filtering.
• Broad LLM Support: Integrates with over 100 LLM providers via LiteLLM, including OpenAI, Anthropic, and AWS Bedrock.
• Developer Experience: Includes a Python SDK and automatic tool conversion for LangChain integration.

Maintenance & Community

The project is hosted on GitHub under the redis organization. Specific details regarding active maintainers, community channels (like Discord/Slack), or sponsorships are not detailed in the provided README.

Licensing & Compatibility

The project is licensed under the Apache License 2.0. This permissive license is generally suitable for commercial use and integration into closed-source applications.

Limitations & Caveats

A separate Redis instance is required for operation. The core AI features necessitate configuration of LLM provider API keys. Production deployments involve managing multiple containerized services (API, worker, Redis).