How do I connect Glen to Goose?

Glen is a standard MCP server, so you add it as a Goose extension (remote MCP) and authenticate over OAuth or an API key. From then on any Goose session can call it to read context and write observations.

Does Glen replace Goose's file-based context?

No. Goose can still read project files; Glen adds durable, org-shared memory that persists across sessions and machines and is also readable by your other MCP clients.

Shared memory for Goose

Goose, Block's open-source AI agent, is built around extensions and speaks MCP natively, so it can pick up tools, run multi-step work on your machine, and act across your stack. What it does not carry on its own is durable, shared memory: each session starts fresh, and anything one Goose run learns about your codebase, your customers, or your conventions evaporates unless you paste it back in or it happens to be sitting in a file. Glen, shared memory for AI agents, gives Goose a long-term memory layer as a single MCP extension, so every session reads from and writes to one org-scoped store instead of starting from zero.

Add Glen as an extension and Goose gets one tool that both retrieves relevant long-term context and records new facts in the same call. Early in a session Goose can pull what your team already knows, the deploy quirk, the customer's constraints, the decision you made last month, and at the end it can write back what this session discovered. You stop re-explaining the same context every time you start Goose, and you stop losing hard-won knowledge between runs.

Because Glen is org-scoped rather than tied to one machine or one session, the memory is shared across every Goose instance on your team and across other agents entirely. A run on your laptop learns something durable; a teammate's Goose, or a scheduled headless Goose run, reads it next time. That is the part local files and per-session context do not give you: persistence that outlives the session and is shared across people and agents. And because Glen is a standard MCP server, the same memory Goose writes is readable from Claude Code, Cursor, or any other MCP client, so whether the work is driven by Goose or by a human in an editor, both draw on the same accumulated knowledge. You wire it in once over OAuth or an API key and let the memory compound.

FAQ

How do I connect Glen to Goose?: Glen is a standard MCP server, so you add it as a Goose extension (remote MCP) and authenticate over OAuth or an API key. From then on any Goose session can call it to read context and write observations.
Does Glen replace Goose's file-based context?: No. Goose can still read project files; Glen adds durable, org-shared memory that persists across sessions and machines and is also readable by your other MCP clients.