Skip to main content
Adapt an AI agent that spends from a vault to your own rules today by changing the hand-written guides, since the AI agent template has no published code to customize yet.
In development. This template is not published yet and details may change. Follow progress on GitHub. Until it ships, use the hand-written guide linked on this page.

​
Why there’s nothing to customize yet

The AI agent template is in development. This page will list its settings once it’s published. The choices below apply to any agent you build from the guides now.

​
Do it today

  • Set the daily limit. Pick the most you’d accept losing in one day if the agent’s key leaked or the model misbehaved. See Spending limits.
  • Choose recipients and tokens. Allow only the addresses and tokens the agent needs. See Agent vaults.
  • Set the approval threshold. Decide which actions a person must approve. See Human approval.
  • Add or remove tools. Every tool you add is something the model can try. Give each one a narrow schema and simulate before sending. See Tool calling.
  • Check trust. If you rely on an identity or reputation contract, plug it in through a small interface. See Identity hooks.
On mainnet an agent spends real funds. Use a separate owner and operator, keep the daily limit small, and complete the security checklist first.

​
Next steps

Security checklist

Check your setup before you fund it.

Templates overview

The templates you can use now.
Last modified on October 2, 2026