# Using Memo (/en/docs/memo/usage)

This page shows what a normal day with Memo looks like once it is installed: bringing in what
changed, finding something, preparing for a conversation and saving what you learned. Each step
links to the page with the details. Most of the time you ask your agent and it runs the commands;
the commands are here so you can check what it does.

## Bring in what changed [#bring-in-what-changed]

Memo reads your folders and mailboxes when you ask, not all the time. One command brings in
everything that changed since the last run:

```sh
memo import
```

Only changed files and new mail are read, so a second run is quick. On Linux, `memo auto on` runs it
for you every few minutes ([keeping Memo up to date](/llms.mdx/docs/memo/archive/content.md)). Telegram and MAX messages come in
through tg and max, not through Memo.

## Find something [#find-something]

> ```text prompt
> Use memo CLI to find what I wrote or received about the venue for the Harbour project.
> Show the matching text and the source of each result.
> ```

Your agent searches your notes, mail and messages together and shows where each result comes from.
When you know where something is, search only there:

```sh
memo search all 'venue'               # messages, mail and notes together
memo search notes 'venue'             # notes only
memo search mail 'from:kai invoice'   # mail only
```

[Search](/llms.mdx/docs/memo/search/content.md) explains the search words and how to give your agent the sources.

## Prepare for a conversation [#prepare-for-a-conversation]

> ```text prompt
> Use memo CLI to show what I know about Kai Sample on Telegram: recent messages, mail and my notes.
> Summarise what is still open between us.
> ```

`memo context telegram:"Kai Sample"` gathers the person's recent messages, mail and the notes linked
to them, each with its source. See [People and projects](/llms.mdx/docs/memo/people/content.md).

## Save what you learned [#save-what-you-learned]

```sh
memo notes add 'Prefers calls in the morning' --about telegram:'Kai Sample'
memo tags add follow-up --note note:NOTE_ID
memo tasks add --project HRB --title 'Book the venue'
```

A note keeps your own words about a person, a message or a project ([notes](/llms.mdx/docs/memo/notes/content.md#write-your-own-notes)).
A tag is a label that helps you find things again ([tags](/llms.mdx/docs/memo/tags/content.md)). A task is something to do
([tasks](/llms.mdx/docs/memo/tasks/content.md)).

## Review what your agent suggested [#review-what-your-agent-suggested]

An agent can propose an action, such as a reply. An app connected through MCP can also save
memories and facts. These wait for you:

> ```text prompt
> Use memo CLI to show the memories, facts and actions waiting for my review, with their evidence.
> Do not confirm or approve anything; I will decide.
> ```

[Memories and facts](/llms.mdx/docs/memo/memories/content.md) explains how to confirm or reject them. Approving an action makes
it a task for you; Memo never sends anything. A memory or fact added with the `memo` command is
confirmed at once. Ask an agent with a terminal to show you the text first, and to add it only after
you agree.

## Answers for scripts and agents [#answers-for-scripts-and-agents]

Add `--json` to most commands to get an answer a program can read. The text answer is for people;
agents and scripts should ask for JSON.

Every result names its source with a **reference**, such as `note:12`, `document:3` or
`msg:telegram/1/77/42`. You can pass a reference to other commands: to tag it, write a note about
it or use it as evidence. [How Memo works](/llms.mdx/docs/memo/how-it-works/content.md#references) explains each kind.

## Choosing an account [#choosing-an-account]

When you have several Telegram, MAX or mail accounts, Memo does not choose one for you. Commands
that work in one account ask for it with `--provider` and `--account`, and tell you which accounts
exist when it is missing. A search with no account searches all of them and says so.