Memory

Sidebar → Memory Profiles (under Tasks), then enable it per-agent in the agent edit page → Memory.

Memory is the agent's automatic long-term memory. A background MemoryLLM reviews each conversation after it goes idle and distills durable knowledge — user preferences, project facts, decisions, lessons learned — into memory entries. On later turns the agent reads them back via the search_memory and read_memory tools.

Think of it as the agent learning from every conversation without you having to teach it explicitly — and without bloating the system prompt.

How It Works

1. Extract — after a conversation idles, the writer model silently reviews the exchange and writes new memories (or updates/removes existing ones).
2. Read — on subsequent turns the agent calls:
   • search_memory — fuzzy/keyword/semantic lookup across stored memories
   • read_memory — fetch the full body of a memory by its slug
3. Maintain — background jobs keep the store healthy:
   • Consolidate — merges and de-duplicates related memories
   • Reconcile — re-indexes and prunes stale entries
4. Delete — removed memories are moved to .archive/ and can be recovered.

Configuration

A Memory Profile defines how memories are extracted and read. Sidebar → Memory Profiles → New:

Field	Description
Name	Display name for this profile
Writer Model	The MemoryLLM used to extract memories (a reasoning-capable model recommended)
Writer System Prompt	Controls what gets extracted
Read Path Template	How retrieved memories are formatted back into the prompt

Then, in an agent → Memory section, toggle Memory on and pick the profile. From the Memory Profiles page you can also Run Now (force extraction), Consolidate, Reconcile, and View Memories.

Memory vs Notes vs Wiki

	Who writes?	When written?
Memory	Background MemoryLLM	After the conversation idles, automatically
Notes	The active agent	Mid-turn, via tool call when relevant
Wiki	Humans (mostly) + agent	Curated; agent can also create/edit

Memory is the most "set and forget" of the three.