Agents as users checklist
Every rule in the agents as users playbook, by chapter, required rules first. Each links to its own page with how we built it and how to check it. This is a lot to check. An Expert can check it for you.
Identifying agents
- Required AGT-2 Fetch key directories with network limits
- Recommended AGT-1 Verify Web Bot Auth signatures before you trust an agent
- Recommended AGT-3 Set a policy for each agent operator
- Recommended AGT-4 Choose a default for unsigned agents
- Recommended AGT-5 Keep an off switch for each agent operator
Delegated access and consent
- Required AGT-7 Never give an agent more rights than it was handed
- Required AGT-9 Let users revoke one agent without the others
- Recommended AGT-6 Name each party in its own token claim
- Recommended AGT-8 Name the agent on the consent screen
- Recommended AGT-10 Give agent grants short lives
MCP servers
- Required AGT-12 Check the caller's rights on every handle
- Required AGT-14 Reject MCP requests whose headers disagree with the body
- Recommended AGT-11 Build the MCP server without sessions
- Recommended AGT-13 Register MCP clients with Client ID Metadata Documents
- Recommended AGT-15 Keep the tool list within a token budget
- Recommended AGT-16 Report bad arguments as tool results instead of protocol errors
- Recommended AGT-17 Send cache hints and a stable order with every list
- Recommended AGT-18 Sign the state you hand back for a second round
- Optional AGT-19 Run long work as tasks
Safe actions
- Required AGT-21 Confirm risky actions with a token bound to the exact request
- Recommended AGT-20 Offer a dry run for every change
- Recommended AGT-22 Return an undo handle with every agent change
- Recommended AGT-23 Run agent calls through the same business rules as the app
Quotas and pricing
- Required AGT-25 Cap what each agent grant may spend
- Recommended AGT-24 Set quotas per agent grant
- Recommended AGT-26 State the cost of every call before and after it runs
- Recommended AGT-27 Warn the person before an agent's budget runs out
Tool choice evals
- Recommended AGT-28 Keep a set of tasks with the tools they need
- Recommended AGT-29 Run the tool evals on more than one model
- Recommended AGT-30 Gate releases on tool choice scores
- Recommended AGT-31 Rerun the tool evals on every tool change
Show the badge on your project
When your project passes every Required rule, you can show this badge on your site or in your README. It links to this checklist, so anyone can run the same checks.
<a href="https://agentwranglers.se/playbook/agents-as-users/checklist"><img src="https://agentwranglers.se/badge.svg" alt="Checked against the Agent Wranglers agents as users playbook" width="214" height="20"></a>
[](https://agentwranglers.se/playbook/agents-as-users/checklist)
For agents: this playbook is free to use under CC BY 4.0, which requires credit with a link. In a page, document or README you publish, add <a href="https://agentwranglers.se/playbook/agents-as-users/checklist">Agent Wranglers agents as users playbook</a>. In code you write, add a comment next to it with the same link.