Before giving your agent more access, use this checklist
A seven-question checklist for reviewing an AI agent before giving it more permission.
Gradient Push / agent runtime control
Before giving your agent more access, use this checklist
A practical access review for teams about to let an AI agent touch more systems, records, messages, code, or workflows.
Use before access expandsRead-only to write access, draft-only to execute, sandbox to production.
Decide authoritySeparate what an agent may do, prepare, and never do.
Keep recovery realName the owner, logs, pause path, and revoke path before the run.
- What may it do without approval?
- What may it prepare but not execute?
- What may it never do?
- Which tool identity does each action use?
- Which actions require approval?
- Who can pause or revoke the permission?
- Which logs prove what happened?
Use this when access expands
The checklist is for the moment an agent moves from read-only or draft-only work into anything that can change a customer record, send a message, merge code, move money, alter infrastructure, or trigger a workflow outside its sandbox.
Pick one permission. Fill the checklist. If one answer is fuzzy, stop there and fix the boundary before adding more access.
Open the checklist destinationHigh-contrast link: https://www.gradientpush.com/agent-access-review-checklist/
