Beta
Agent spending policies
Last updated
TL;DR: An agent owner sets spending rules (currency, per-purchase cap, daily cap, allowed merchants and categories) on the agent's credential. Each proposed purchase gets a deterministic, signed allow or deny receipt that merchants verify free. Rightshold never holds or moves money.
Set daily and per-purchase caps, a currency, allowed merchants and categories for an agent credential. Every proposed purchase gets a deterministic, signed allow or deny receipt merchants verify free.
What it is
The owner of an agent sets a spending policy on the agent's Rightshold credential: a currency, a per-purchase cap, a daily cap, the merchant domains and categories it may buy from, and an optional expiry. Before the agent pays, it asks Rightshold to authorize the purchase. The answer is allow or deny with the exact reasons, signed with Rightshold's issuer key and written to the public ledger (stream spend_authorizations).
How the decision is made
Plain code, same inputs, same answer. A purchase is denied when the credential is revoked, there is no policy, the policy has expired, the currency differs, the amount is over the per-purchase cap, today's allowed total (UTC day) plus this amount is over the daily cap, the merchant domain is not on the list (*.domain allows subdomains), or the category is not on the list. No AI model is consulted.
What it does not do
Rightshold never holds, moves or settles money. It records the owner's rules and signs each decision. Merchants should accept an allow receipt only before its valid_until time (15 minutes).
Use it
- Set a policy (owner account key): POST https://www.rightshold.com/v1/spend/policies with {credential_id, currency, per_tx_max_minor, daily_max_minor, allowed_domains, allowed_categories, expires_at}
- Authorize: POST https://www.rightshold.com/v1/spend/authorize with {credential_id, amount_minor, currency, merchant_domain, category, reference}
- Running total: GET https://www.rightshold.com/v1/spend/policies/{credential_id}
- Merchants verify free: POST https://www.rightshold.com/v1/spend/verify with {receipt, signature}, or GET https://www.rightshold.com/v1/spend/authorizations/{id}
- MCP: set_spend_policy, authorize_spend, verify_spend_authorization (https://www.rightshold.com/mcp?toolset=trust)
Price
Checking and verifying are free. Each signed authorization is planned as a paid line, announced at /transparency/changes with 7 days' notice before it joins the pricing rulebook. Free during the preview; Rightshold credits have no cash value during the preview.
Frequently asked questions
How does a merchant know an agent purchase is within its owner's limits?
The agent sends the purchase to POST /v1/spend/authorize first and hands the merchant the signed allow receipt; the merchant verifies it free and accepts it only before its valid_until time (15 minutes).
Does Rightshold hold or move the agent's money?
No. It records the owner's rules and signs each allow or deny decision; payment happens wherever the agent already pays.
Is the decision made by AI?
No. Plain code applies the published rules, so the same inputs always give the same answer and every denial lists its exact reasons.