How is Glen different from Mastra's working and thread memory?

Mastra's memory keeps a single agent coherent across its own threads. Glen is durable, org-shared long-term memory that outlives any thread and is read and written by every agent in your organization, including agents outside Mastra.

How do my Mastra agents reach Glen?

Glen is a standard MCP server. Connect it through Mastra's MCP client support and call it from any agent or workflow step to retrieve context and record observations in one round trip.

Shared memory for Mastra

Mastra gives TypeScript developers a clean way to build agents, workflows, and tools, with working memory and a thread-based memory layer baked in for conversational recall. That memory is excellent for keeping a single agent coherent across a thread, summarizing what was said and recalling it later. But it is scoped to the agent and its threads: the durable knowledge your whole organization's agents should accumulate, across deployments, across different Mastra agents, and across other frameworks entirely, is not something a per-thread memory store is meant to hold. Glen, shared memory for AI agents, fills that gap with one MCP tool every agent in your org reads from and writes to.

Wire Glen into your Mastra agents as an MCP tool through Mastra's MCP client support and each agent gains a single capability that retrieves long-term context and records new facts in one round trip. Inside a workflow step or a tool call, an agent can pull what the organization already knows before it acts, then write back the durable facts the run produced, no extra persistence layer, embedding pipeline, or write path to maintain alongside Mastra's own memory.

The distinction matters: Mastra's working and thread memory keep one agent coherent within its conversations, while Glen is the org-wide layer that outlives any thread and is shared across every agent. A fact one Mastra agent records is available to a different Mastra agent in another service, and to agents running in LangGraph, CrewAI, or a coding tool, because Glen is a standard MCP server rather than a TypeScript-only library. The knowledge compounds across your whole fleet instead of staying inside one agent's threads. Connect once over OAuth or an API key, and the same memory your Mastra agents build is readable from Claude Code, Cursor, or any other MCP client, so automated and human-driven work draw on one source of truth.

FAQ

How is Glen different from Mastra's working and thread memory?: Mastra's memory keeps a single agent coherent across its own threads. Glen is durable, org-shared long-term memory that outlives any thread and is read and written by every agent in your organization, including agents outside Mastra.
How do my Mastra agents reach Glen?: Glen is a standard MCP server. Connect it through Mastra's MCP client support and call it from any agent or workflow step to retrieve context and record observations in one round trip.