Hive
Provides shared persistent memory for AI agents through an MCP-compatible server.
AI agents often operate as stateless entities, losing context with each new interaction. Hive addresses this by offering a persistent, shared memory store, enabling agents to retain and recall information across conversations. It functions as an MCP server, allowing any MCP-compatible client like Claude Code, Cursor, or custom agents to store and retrieve memories by key or tag, all managed through an intuitive web user interface. A free hosted service is available, simplifying setup and requiring no AWS account.
Key Features
01Web UI for memory management and agent access control
02Synthesize memories on a topic into a summary (`summarize_context`)
03Durable and shared memory for AI agents
04Store and retrieve memories by key or tag (e.g., `remember`, `recall`, `list_memories`)
05OAuth-based authentication for tracking agent/user contributions
060 GitHub stars
Use Cases
01Enabling multiple AI agents or team members to share and manage a common memory store
02Providing persistent memory to stateless AI agents across conversations
03Connecting various MCP-compatible clients (e.g., Claude Code, Cursor) to a central memory service