# Access and safety

What your agent can read and change in myPip, what stays with you, and how to switch it off.

Your agent acts as you. It can do what you allow when you connect it, and nothing you can't do
yourself in myPip.

## What you approve

When you connect, myPip shows what the agent is asking for. You choose.

- **Read.** Your agent sees your people, why each is ready, your brief, your plan and what's new. It changes nothing.
- **Reach out.** It takes a way in, reports what it sent, adds what it found and answers a card. These change things in myPip only.
- **Manage.** It can make itself the way I reach you. That's one setting.

Read is always on. The others are yours to leave off.

Your agent can't change your plan, your card or your password. Those stay in myPip.

## What stays with you

Your Reddit, LinkedIn, X and email logins live in your agent. I never see them and I never send
from those accounts. Your agent does, as you.

That means the risk on those accounts is yours. Platforms limit automated sending, LinkedIn most
of all. Keep it to a few a day. I only ever hand over one way in per person.

## What I won't ask your agent to do

> [!pip-wont]
> Find out who is behind an anonymous account. Send to the same person twice. Write to anyone who
> said no.

When I can't find something about a person, I tell your agent where to look, and only in what that
person has published about their work.

## Email addresses

If a person or their company published a work address, your agent gets it, along with the page it
was published on. It's the same address you see in myPip.

If the only address I have is one I checked through a data provider, your agent is told I can
email that person, and not the address. Those I keep to myself.

An address I won't write to stays held back too: a personal inbox, a front desk, a bounce, or
anyone who said no.

## When an introduction can't be made

If you're out of introductions, your agent is told plainly, and that nothing was sent. It isn't
pointed at a plan or a price.

## Words from the web are not instructions

Some of what your agent reads through me was written by strangers: posts, comments, profiles. I
mark all of it as untrusted, and tell your agent to treat it as information about a person, never
as something to act on. The way in and the rules are written by me.

## See what's connected

**Profile** in myPip lists each connected agent under *Your agent*, with what it's allowed to do.

## Switch it off

Press **Disconnect** next to the agent in Profile. It stops working straight away. You can also
remove Pip from the agent's own connector list.

If your agent goes quiet for 48 hours while it's the way I reach you, I go back to your usual
channel on my own.
