The plan
You approveBefore any code exists, you read the plan. A second AI (Fammy, our parrot) has already checked it for gaps. Approve it, change it, or send it back.
Security: what the AI can and can’t do
Letting an AI work on your project raises fair questions. Can it break something? Where does my code go? Who can see my login? Here are the straight answers.
Explained for people with an idea.
Can it do something dangerous without asking?
No. It stops in three places and waits for you.
Risky actions
It asksDeleting things for good, using admin rights, pushing code, publishing, or connecting to other machines: it stops and asks. Allow it once, or say no with a note.
For example rm -rf, sudo, git push and npm publish. Editing a .env file asks too.
Going live
You approveYou see a preview and what changed. Ship it, send it back with a sentence, or don’t ship.
Only a project lead can connect a Claude login or publish to production.
On Builder and Fleet, the AI works inside a sealed-off container.
Will it write insecure code?
Every time the AI finishes a piece of work, security scanners check it before it counts as done. The AI doesn’t grade its own work.
After every build step
Scanners look for
- Passwords and keys left in the code
- Known security holes in code it added
- Common ways attackers break in
- Setup mistakes, like a door left open
Nothing found
The step is done
The work moves on. Nothing for you to do.
Something found
The step stops and tells you
You get a list of each problem: where it is, how serious it is and how to fix it. Tap Fix and the AI fixes it.
- You choose how strict. Set the level for each project.
- Your code stays put. Only the names and versions of added code libraries go out, to check them against known problems.
- Exceptions are written down. A project lead can accept a risk, and the reason is recorded.
Can it touch the rest of my computer?
No. When it works on your computer, it sees your project and nothing else.
Only your project folder.
It reads the files of the project it works on. Your documents, photos and other projects are out of its reach.
Secret files stay shut.
Files that hold passwords and keys are off limits, even inside your project.
Your own Claude’s extras stay yours.
If your own Claude can reach your email, your bank or other tools, the AI working for ShipItFam can’t. It starts with none of them.
Commands run in a sealed box.
When it runs a command, it runs in a sealed box that holds only its copy of your project.
Only the jobs you approved.
When your Kraken (the ShipItFam app on your computer) starts work, it takes the jobs you approved and nothing else. A job someone else sends you waits for you.
Other people’s words are read, not obeyed.
A message from a teammate can’t give orders to your AI. ShipItFam marks other people’s words as something to read, never instructions to follow.
If your own Claude can pay or send mail for you, give it a separate Claude profile from the one you code with.
Where does my code go?
Into a repository you own, and nowhere else until you say so.
Separate copies for each job.
Each job works on its own copy of your project, so your live version is never touched while it works.
You choose when code leaves.
When you connect a repository, it starts on “ask me first”, so it checks with you before sending anything.
No repository yet? Nothing leaves.
Start from a ready-made project and connect a repository later, when you want to.
Your Git password stays hidden.
The access key for your code host never reaches the AI or the commands it runs.
Going live has its own button.
It asks you to confirm, and one tap undoes it.
Leaving is easy.
Your code stays where it is. Cancel, and the repository is exactly as the AI left it.
Who can see my Claude login?
ShipItFam runs on Claude, an AI from Anthropic, signed in as you. You pay Anthropic for Claude and us one flat fee. We don’t resell Claude and we don’t use a hidden account of our own.
You sign in once.
Approve a link and paste the code back. No API keys to copy around.
On Free.
Your login never leaves your computer.
On Builder and Fleet.
Only your own projects’ crews use it. It is never resold and never written to logs.
Remove it any time.
Sign out from the app and it is gone.
It is the real Claude Code login, the one claude login creates. No long-lived token in an environment variable.
What do you keep about me?
Only what the app needs to show you your work.
Your account.
Your email address, from signing in.
Your jobs.
What you asked for, the plan, a summary of each step, what changed and the live log, so you can follow along from your phone.
Your code.
It lives in your repository. The AI works on a copy: in its own workspace on Builder and Fleet, on your computer on Free.
Your conversations with Claude.
They go to Anthropic through your own login, under Anthropic’s terms.
Never.
We don’t sell your data, and we don’t train any AI on your code.
Want a copy of your data, or want it deleted? Email privacy@shipitfam.com.
Found a security problem? Tell us on the contact form and pick “Security or privacy”.
Still have a security question?
Ask us. Someone who builds ShipItFam will answer.