What the agent remembers and how

How You don't have to think about files. The agent does. But there are three habits about its memory worth knowing.

Save it, don't just show it

A common beginner mistake: dropping a document into the chat and assuming the agent now "has it forever." Wrong. The contents only live in the current conversation. The moment the chat resets or rolls over, the document is gone unless the agent saved it somewhere durable. The right move: when you share something you'll need again, tell the agent — "save this in your memory so you always have access to it." The agent stores it in a place that survives every session.

Project folders

For any real piece of work, ask the agent to set up a project — its own space with a plan and a running log. A rule worth adding to always-loaded context: "before working on any project, read the plan and the running log first." That one rule means the agent picks up where it left off, even if you come back to the project a month later.

Send it the error

When something breaks, you do not need to read logs. Copy the error, screenshot the weirdness, paste it into chat. The agent reads its own errors better than you can.

Don't try to be organized for it

Resist the urge to over-organize on the agent's behalf. The agent's own memory and search work fine on plain structures. Spend that energy on better rules and skills instead. Three small habits about the agent's memory worth knowing, so nothing important slips through.