Memory

The Investigator remembers. When it learns something worth keeping — a fact about your infrastructure, a convention your team follows, a preference you’ve stated — it can save that as memory and reuse it in future conversations, so you don’t re-explain your setup every time. Microtica Agents keeps memory alongside skills as the customization layer of the product.

How memory gets written

Memories are usually saved from chat: the Investigator writes them down during a conversation, and you can ask it directly to remember something (“remember that staging shares a database with dev”). You don’t write memory files by hand in the console — you review and prune what the Investigator has kept.

Two scopes

Memory is split the same way skills are:

• Project memory — facts and conventions the Investigator can reuse only in this project. This is where infrastructure knowledge belongs: what services exist, how they’re deployed, what past incidents taught.
• Your preferences — preferences that apply to you across all projects, like how you want findings reported. These override project memory when they conflict.

Review what’s remembered

1. In the console, open your project’s settings and choose Memory under Agent.
2. Each memory appears as a card with its content rendered in full and a Last updated timestamp, grouped under Project memory and Your preferences.

If the project section is empty, the page says so: “No project memory yet. Ask an agent to remember something in chat.”

Remove a memory

Memory is worth auditing occasionally — a stale fact steers investigations the wrong way longer than no fact would.

1. Click the trash icon on the memory’s card.
2. Confirm in the Remove memory? dialog. The Investigator will no longer use it in future conversations.

Removal is the only edit the console offers. To correct a memory rather than delete it, tell the Investigator the updated fact in chat and it can save the correction.

Next: Members & roles