Memory

Memory is what makes Goose a persistent coworker instead of a stateless chatbot. Goose remembers important context across conversations so it can build on past work.

​

How memory works

As you work with Goose, it picks up on important details — your projects, your team’s preferences, key decisions, competitors, past research findings — and stores them in memory. The next time you start a chat, Goose already has that context. Memory is stored as structured files in your workspace:

• MEMORY.md — Long-term learnings and key facts
• Daily session logs — Summaries of what happened each day
• USER.md — Information about you and your preferences
• COMPANY.md — Your business context (industry, product, target market, etc.)

​

What gets remembered

Goose stores things like:

• Key facts about your business, products, and goals
• Research findings and competitive intelligence
• Your preferences and working style
• Decisions made in past sessions
• Project context and team workflows

​

Updating memory

Goose may propose memory updates during a conversation — for example, after learning something important about your business. These proposals are shown to you for approval before being saved. You can also ask Goose to remember specific things: “Remember that we use Linear for task tracking and deploy to Vercel every Friday.”

​

Identity

Your agent also has an identity — a name, avatar, and personality. This shapes how Goose communicates. You can customize your agent’s identity in Settings.