Agent Permission and Risk Review for a Proposed Automation

Assess risks before an agent is given access.

Added October 15, 2026. Written with AI assistance and checked for clarity and placeholders; not reviewed by a lawyer, accountant, clinician or other professional. How these prompts are made

The Open buttons send the prompt text as currently shown below to that site only when you click. Fill in the placeholders first.

[VERIFY] is an editorial flag, not a blank to fill. Wherever the prompt shows [VERIFY], check that item against a reliable source before you use it (what [VERIFY] means).

Review a proposed automation. What it will do and for whom: [AUTOMATION_PURPOSE]. Systems and data it will touch: [SYSTEMS_AND_DATA]. Actions it can take: [ACTIONS]. Existing safeguards: [SAFEGUARDS].

For each action rate impact if wrong (low, medium, high), reversibility, and the safeguard in place. Make a table with Action, Impact, Reversible, Safeguard, Gap, and Proposed control (approval, limit, logging, dry run). Identify actions that should require human approval, data that should be excluded, and permissions that can be narrowed. Count high-impact actions without a safeguard. Provide a go, go with conditions, or hold recommendation with reasons, and a pilot plan.

This is a risk review aid. Mark regulatory or contractual points as [VERIFY] with the responsible team.

Add a worked review of an invoice-approving agent: reads invoices (low impact), flags duplicates (low), approves payments under 100 (medium), changes bank details (high, should be never), with the safeguards and gaps; and a decision of 'go with conditions' listing the conditions.

Fill in before use

Optional: type your details and the prompt above updates. Nothing is saved or sent.

Replace each bracketed placeholder with your own details, then paste into Any AI or another assistant.