Agent Permission Matrix and Least-Privilege Review

Define exactly what each agent may access and change.

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).

Create a permission matrix for the agents or automations at [ORGANISATION_OR_PERSON]. List of agents and their jobs: [AGENT_LIST]. Systems and data they touch: [SYSTEMS_AND_DATA]. Sensitive data classes: [SENSITIVE_DATA].

Build a table with rows for each agent and columns for each system showing the minimum access needed (none, read, write, delete, send, pay), the reason, who approved it, review date, and the credential type used (personal, service account, scoped token). Identify excessive permissions, shared credentials, and agents that can both create and approve the same item. Propose changes to reach least privilege and a plan to rotate credentials. Add an approval workflow for new permissions and a quarterly review checklist. Write a one-page summary for an executive that states the biggest three risks in plain language.

Do not include real secrets. Use descriptions of credentials, and mark anything I should confirm with my security team as [VERIFY].

Add a worked example for a scheduling agent: calendar read yes, calendar write with approval, email read none, email send never, contacts read only the shared directory, and the reason for each. Then state the test: remove each permission one at a time and see if the agent can still do its job.

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.