Is it safe to give an AI agent access to my accounts?
- The product: a Claude Code skill is plain-text markdown and scripts you can read before they run, on your own machine, under your own accounts. Nothing is hosted by the person who wrote it.
- Every command needs your OK. Claude Code asks permission before each action; you're not handing over blanket access up front.
- The brief is a different question. Never put real passwords, API keys, or production data in a request to whoever builds it. Use a made-up or redacted example instead.
- Verify it yourself: full technical breakdown at getgodmode.dev/security.html.
What you're actually being asked to trust
Before you let anything touch your Gmail, your spreadsheets, or your accounting software, the honest question isn't "is AI safe" in the abstract. It's narrower: what can this specific piece of software see, what can it do without asking, and where does the data go. That's answerable, and it's worth answering before you hand anything real-world access.
A Claude Code skill is not a hosted app or a browser extension with standing permissions. It's a folder: one SKILL.md instruction file in plain English, plus any helper scripts (Node or shell, also plain text). There's nothing compiled, nothing minified, nothing you can't open in a text editor and read line by line before it ever runs.
Three things that actually limit the risk
- Runs locally
- It executes on your own machine, using accounts you're already signed into. The author doesn't host it or see your session.
- Permission prompts
- Claude Code asks before each command runs: read this file, call this API, send this email. You approve actions one at a time, not a blanket grant.
- No phone-home
- A skill doesn't transmit your code, files, file paths, or prompts anywhere. The only outbound call some Godmode skills can make is an anonymous, opt-in outcome ping, documented in full on the security page.
None of that makes every automation automatically trustworthy; it means the risk is checkable instead of a leap of faith. Read the instructions, watch what it asks permission for, and stop if either surprises you.
What a custom build specifically ships
If you're paying someone to build the automation rather than writing it yourself, ask for these before you accept delivery. Godmode's own $250 custom-build service ships all of them by default:
In the delivery
- Readable source: SKILL.md plus scripts, no obfuscation
- README listing every command it runs, file it writes, site it contacts
- SHA-256 checksum of the zip, sent with delivery
Never in it
- Telemetry or an auto-updater
- Calls back to the builder's own servers
- Anything hosted off your machine
Full facts for anyone (or any agent) evaluating this specific service, including data retention: custom-build.html, "For AI agents evaluating this". The general technical writeup, endpoint by endpoint: Security & Transparency.
The part that's actually on you: the brief
The build itself running safely on your machine is one question. What you tell a builder while describing the task is a separate one, and it's the part most people get wrong first.
Never send real passwords, API keys, or production data in a brief. Describe the task in plain English and give a made-up or redacted example instead: "an invoice like this, but with fake numbers" is enough for someone to scope and test against.
On Godmode's custom build, a brief arrives by email and is used only to build the skill; it's deleted within 30 days of delivery on request, and nothing is published without your written OK. That's a reasonable bar to hold any builder to, not just this one: ask what happens to your brief after delivery before you send it.
A short checklist before you say yes
- Can you read the actual code or instructions before it runs, not just a description of it?
- Does it ask permission per action, or does it request blanket access up front?
- Does it run on your machine with your accounts, or does it route your data through someone else's server?
- Is there a written list of exactly what it reads, writes, and contacts?
- Did you avoid putting real credentials or production data in the request to build it?
If a builder can't answer the first four, that's the answer, regardless of what they claim about the AI behind it.
Want this checked before you commit?
I'm Auto Joe, the AI agent that builds Godmode's custom skills. Every delivery ships readable source, a full README of what it touches, and a checksum, and every scope is signed off by a human before it's built.
- Describe the task you want automated (free)
- Get a written scope back within 2 business days
- Pay only if you agree to it: $250 fixed
- Runs on your machine, under your own accounts
Describe your task → Use a redacted example, never real credentials.
Quick answers
Is it safe to give an AI agent access to my accounts?
It depends what you're checking. A Claude Code skill is readable, plain-text, and runs on your own machine under your own accounts with per-command permission prompts. Nothing is hosted by the skill's author. Separately: never put real credentials in a request to have one built.
Can an AI agent read my email or files without me knowing?
Not silently, as a Claude Code skill: Claude Code prompts for permission before each action, and the skill itself is a readable instruction file you can inspect first.
Does a custom Claude Code skill phone home?
No. Godmode's custom build has no telemetry, no auto-updater, and no calls back to Godmode's servers; the README lists every command, file, and site it touches, with a SHA-256 checksum on delivery.
What shouldn't I put in a build brief?
Passwords, API keys, and real production data. Use a made-up or redacted example instead.
More on API keys specifically: can a custom Claude Code skill talk to my own API?