Memory

Memory stores long-term facts, preferences, instructions, and summaries that persist across sessions.

Memory kinds

Kind	Use for
preference	User likes and dislikes (e.g., "prefer short bullet answers").
fact	Factual knowledge about the vault or user (e.g., "main project is in /Work/").
instruction	Standing directives (e.g., "always add a date header").
summary	Condensed session takeaways (e.g., "decided to migrate to new folder structure").

Memory scopes

Scope	Visibility
global	All vaults and sessions.
vault	Only this vault.
project	Only the active project (filtered until project context is active).

Commands

Command	What it does
/memory add [kind] [scope] <text>	Store a new memory. Kind and scope default to preference and vault if omitted.
/memory review	Browse stored memories in a searchable list.
/memory manage	Alias for /memory review.
/memory export	Export memories to a vault note.
/memory clear --confirm	Delete all stored memories. Requires confirmation.

Memories are stored in memories.jsonl inside the plugin directory. They are never injected into context automatically; the agent must call the search_memory tool to retrieve them.

When the agent uses memory

The model decides when to search memory based on the conversation. You can prompt it explicitly: "What do you remember about my preferences?"

Privacy

• Memories are local files inside the plugin directory.
• Nothing is sent to providers unless the agent calls search_memory and the result is included in the prompt.
• Clear memories with /memory clear if you want to reset.