The agent permissions review

Before you connect an agent, MCP server or skill: list what it can do, what it reads, and decide what it's allowed to do without you.

Time: About 25 minutes
Version: 1.0, updated 12 Nov 2026
From the guide: Agents, MCP servers and skills: what each is and when to use one

How to use it

1Fill it in from the tool's own documentation, before you connect anything.
2List every action, and mark which ones change things or can't be undone.
3Review it again whenever the tool updates or you add a connection.

Prefer a file?Designed PDF (blank, with example answers)Plain text, for copying into an AI tool

Save my answers in this browser

Answers stay on this device. They're saved only if you tick the box, and only in this browser. Nothing is sent to us. Don't save sensitive details on a shared computer.

1. What I'm connecting

Agent, MCP server or skill

Who made it, and why I trust them

Workflow or agent?

Fixed steps I set (workflow)
AI chooses the steps (agent)

2. Actions

Allow: runs without asking. Confirm: asks me each time. Off: switched off.

What it can do

	Action	Read-only, changes things, or can't be undone	Allow, confirm or off
Action 1
Action 2
Action 3
Action 4

3. Outside content and injection

Outside content it will read (web pages, emails, files from others)

Protections

Nothing that sends, deletes, publishes or spends runs without my confirmation
It has the narrowest access that does the job
I'll read each confirmation prompt, not click through
I know how to disconnect it

Limit what an agent can do, not just what it's told. Anything it can't do, injected text can't make it do.

How to use it

One useful email a week.