Memory
The Memory page lets you browse and manage the entries stored in tota's Second Brain — the SQLite-backed persistent memory that the agent reads before every response.
#Memory types
tota stores two categories of memory, shown as separate sections:
| Section | Description |
|---|---|
| Short-term | Recent conversation context and episode summaries |
| Long-term | Persistent facts extracted automatically — identity, preferences, goals, projects, habits, decisions, constraints, relationships, and reflections |
Each entry shows:
- Content — the stored fact or summary
- Tags — optional labels for filtering and retrieval
- Type — one of the 10 memory types
- Timestamp — when the entry was created or last updated
#Adding a memory entry
Click + Add to open the entry modal. Fill in:
- Content — what the agent should remember (required)
- Tags — comma-separated labels, e.g.
work, project-x(optional)
Click Add to save. The entry is written to the SQLite database immediately and will be available to the agent on its next response.
#Editing an entry
Click the pencil icon on any entry to open it in the edit modal. Update the content or tags and click Update to save changes.
#Deleting an entry
Click the trash icon on any entry. The entry is removed from the database immediately. Deletion is permanent.
#Relationship to /memory in chat
Changes you make in the Memory page are equivalent to what the agent does via the /memory CLI command. You can use either interface — they write to the same database.
#Auto-extraction
tota extracts 0–3 new memory facts automatically after each response. These appear in the Memory page without any manual action. See Second Brain for the full extraction and consolidation rules.