Ontario Protocol.
x402 pre-payment decision playbook

Run agent can-pay preflight for MCP tool builders

The agent needs an allow, review, or deny decision before wallet spend. This playbook maps the decision to observable Ontario evidence so an agent can fail closed before payment.

Run can-pay preflight MCP builders No-payment first
Decision question: Can MCP tool builders rely on this x402 endpoint without paid tools that agents cannot inspect before invocation?

Only after MCP tool builders can inspect target URL, max price, network, report freshness, and policy outcome. If the evidence is missing, stale, or contradictory, the agent should fail closed.

Required evidence

  • Confirm the endpoint maps to a real payment resource for builders exposing paid MCP tools or HTTP tools to agent runtimes.
  • Require target URL, max price, network, report freshness, and policy outcome.
  • Reject when there is policy deny, stale readiness evidence, unsupported network, or price above the agent limit.
  • Link the final decision to a public report, manifest, benchmark row, or policy outcome.

Reject conditions

Reject or send to human review when there is policy deny, stale readiness evidence, unsupported network, or price above the agent limit.

Do not let a positive-looking badge or directory row override a stale or contradictory live check.

Agent workflow

  1. Start with the no-payment evidence path.
  2. Run the Ontario verifier or policy endpoint shown below.
  3. Compare live endpoint behavior against manifest and discovery metadata.
  4. Store the report_id or policy decision with the agent's payment log.
  5. Send the payment only when the agent's own policy allows it.

Command or probe

curl -X POST https://ontarioprotocol.com/api/agent/can-pay \
  -H "Content-Type: application/json" \
  -d '{"target_url":"https://example.com/api/paid-endpoint","max_price_usdc":"0.05"}'

Primary Ontario surface: /api/agent/can-pay

Sourceability note

Search Console showed early impressions for /answers/agent-payment-firewall.

KPI: more paid tools with machine-readable verification metadata.

Trust boundary

Ontario Protocol provides readiness and trust signals based on observable endpoint behavior, manifest metadata, payment challenge behavior, report history, and public evidence. It does not guarantee endpoint safety, service quality, financial outcome, or future behavior.

Use the agent payment preflight guide Run verifier Browse discovery Read trust boundary