Shared memory for Dify

Dify lets you build LLM apps and agents visually, chaining prompts, tools, and knowledge-base retrieval into workflows without writing much code. Its knowledge bases are great for grounding answers in documents you upload, and conversation variables carry state within a session. But a knowledge base is a curated document store you maintain by hand, and conversation memory ends when the session does. The durable, growing knowledge your Dify apps should accumulate from their own runs, and share with agents outside Dify, is a different thing entirely. Glen, shared memory for AI agents, provides that org-wide memory as a single MCP tool your Dify apps read from and write to.

Add Glen as an MCP tool in your Dify workflow and any app or agent node gains one capability that retrieves long-term context and records new facts in a single round trip. Early in a flow, a node can pull what the organization already knows about the user or topic to ground the response; later, a node writes back the durable facts the run produced, so your app gets smarter automatically rather than waiting for you to manually update a knowledge base.

That is the difference from Dify's built-in retrieval: a knowledge base holds documents you put there, while Glen holds the facts your agents learn while running, and it keeps them across sessions and across apps. Because the memory is org-scoped, a fact one Dify app records is available to every other Dify app in your workspace, and to agents running in other frameworks or coding tools, because Glen is a standard MCP server rather than a Dify-internal feature. The knowledge compounds with every conversation instead of resetting when the session closes. Connect once over OAuth or an API key, and the same store your Dify apps build is readable from Claude Code, Cursor, or any other MCP client, so your no-code apps and your developers' agents share one source of truth.

FAQ

How is Glen different from a Dify knowledge base?
A Dify knowledge base holds documents you curate and upload for retrieval. Glen holds the durable facts your agents learn while running, shared org-wide and updated automatically, and readable from any MCP client, not just Dify.
How do my Dify apps reach Glen?
Glen is a standard MCP server. Add it as an MCP tool in your Dify workflow and any node can call it to retrieve context and record observations in one round trip.