What's the difference between /private/, /workspace/, and /workspace/teams/<slug>/ paths?
Understand the three memory scopes in Unison: private (caller-only), workspace (all members), and team (a sub-group).
What path should I write to — /private/, /workspace/, or /workspace/teams/<slug>/?
Unison organizes memory into three scopes. The path you write to determines who can read it back.
The three scopes
/private/
Only the caller (or, in multi-user apps, the actor whose ID is on the request) can read this content. Use it for personal notes, user-specific preferences, or anything that should never cross user boundaries.
# Write a private note
PUT /v1/brain/doc # body: {"path":"/private/my-note.md","bodyMd":"...","title":"..."}/workspace/
All authenticated members of the workspace can read this. Use it for shared company or project knowledge — architecture decisions, team processes, shared glossaries.
# Write a workspace-wide fact
PUT /v1/brain/doc # body: {"path":"/workspace/architecture/db-choice.md","bodyMd":"...","title":"..."}/workspace/teams/<slug>/
A sub-group within the workspace. Only members of that team can read content here. Use it when you want shared memory within a team but not visible to the whole workspace.
# Write to a specific team's namespace
PUT /v1/brain/doc # body: {"path":"/workspace/teams/platform/sprint-goals.md","bodyMd":"...","title":"..."}Quick decision guide
| What you're storing | Path |
|---|---|
| Personal notes / user-specific memory | /private/ |
| Company-wide knowledge, shared docs | /workspace/ |
| Team-internal context | /workspace/teams/<slug>/ |
Recall respects these scopes
When you call GET /v1/brain/context, the brain automatically searches across all scopes the caller has access to. A workspace member gets results from /workspace/ and their own /private/. Use pathPrefix to restrict to one scope when needed.
How do I limit recall to a specific folder or team?
Use pathPrefix to scope Unison recall to a particular path, preventing irrelevant sections of the brain from diluting results.
How does Unison extract facts from what I ingest?
Understand how Unison's async entity-resolution and fact-extraction pipeline turns raw conversations into structured knowledge.