Memory and Triggers

Memory & Triggers

Persistent memory

Real Chat reads a memory note on every message, so facts you want it to always know stay in context across sessions and restarts. The path is set in Settings (it's shown in the in-app help, e.g. a file under _memory).

• Say "remember this: …" and the model appends it to the memory note.
• Edit the memory note directly any time to add, change or prune facts.
• Memory counts toward your context budget — the ctx ~Xk indicator includes it. If it grows large, trim it (see Usage → Context).

Triggers — your own commands

Inside the memory note, add a ## Triggers section with natural-language commands. Each trigger maps a phrase to an action. For example:

## Triggers
- **add book**: when I say "add book <title>", append "- <title>" to "Reading/Reading list.md".

Then typing:

add book Dune

…runs that instruction — here, appending - Dune to your reading list. Triggers are just instructions the model follows, so they can do anything the model's tools allow (append, create, edit, search, etc.).

Tips

• Keep triggers short and unambiguous; name the exact target file in quotes.
• Use memory for durable preferences ("always answer concisely", "my projects live in Work/") and triggers for repeatable actions.
• Because triggers can edit files, consider leaving YOLO mode off so you confirm changes — or on, if you trust your triggers and want them instant.