Verify an x402 endpoint before an agent pays.
Ontario checks whether a paid endpoint is readable, challenge-correct, discoverable, and ready for an agent's task before a wallet signs.
Providers can also test the full purchase path for one real buyer task: HTTP 402 readiness, catalog activation, five live Bazaar searches, budget and network fit, and public demand evidence. Pay only when the result identifies a qualified repair worth buying.
No wallet, payment, private repository, or API key. Public receipts are created only with explicit opt-in.
Provider buyer path
find_first_blockerFind the first purchase blocker
Submit one public endpoint and the exact task a funded agent should buy it for.
Free and read-only. No wallet, payment, private data, or paid tool invocation. The result recommends paid work only for a qualified, observable gap.
# 1. Verify whether an endpoint is x402/Bazaar ready (free) curl -X POST https://ontarioprotocol.com/api/verify/x402-readiness \ -H "Content-Type: application/json" \ -d '{"target_url":"https://example.com/api/paid-endpoint"}' # 2. Load the agent trust policy (free) curl https://ontarioprotocol.com/.well-known/x402-trust.json # 3. Discover endpoints with grade=ready (free) curl "https://ontarioprotocol.com/discover?grade=ready" # 4. Discover everything Ontario exposes (free) curl https://ontarioprotocol.com/.well-known/x402.json # 5. Try the free live trust preview — no wallet, no payment, no evidence bundle curl -X POST https://ontarioprotocol.com/api/demo/agent-trust-scan/live \ -H "Content-Type: application/json" \ -d '{"target_url":"https://example.com"}' # 6. MCP hosts can call agent_trust_preview with the same target_url before selecting the paid scan. # 7. Probe the paid endpoint without paying — get back HTTP 402 curl -i -X POST https://ontarioprotocol.com/api/x402/agent-trust-scan \ -H "Content-Type: application/json" \ -d '{"agent_id":"my-agent","target_url":"https://example.com"}' # Response: HTTP/1.1 402 Payment Required PAYMENT-REQUIRED: <base64-encoded PaymentRequired JSON> WWW-Authenticate: x402 # 8. Sign + retry with PAYMENT-SIGNATURE header (using x402 client SDK). # 9. Get back: HTTP 200 + structured trust report + settlement receipt when available.
For agents: paid calls stay opt-in
Agents should use the free verifier, can-pay policy, and strict
discovery catalog before signing anything. If a paid trust scan is
still justified, the walkthrough shows the exact
PAYMENT-SIGNATURE retry path.
MCP hosts should call agent_trust_preview first and pay only when the preview is insufficient and their own spend policy allows it.
The shortest provider path to agent discovery
Ontario's provider path is deliberately simple: prove the public contract first, then pay only for the directory submission you actually need. A listing is evidence and distribution metadata, not a promise of traffic, ranking, safety, or purchases.
1. Score readiness free →
Check one public endpoint, repository, and MCP surface before spending. Keep the scorecard and report ID.
2. Submit the ready endpoint →
Validate the payload for free, then use the 0.50 USDC x402 listing route when the public evidence is aligned.
3. Need assisted publication? →
Use the optional outcome-based pilot only when the endpoint is ready but agreed catalog surfaces still need provider-controlled publication.
Why aren't agents buying? →
Find the first blocker across payment readiness, Bazaar activation, buyer-task retrieval, budget and network policy, and public demand evidence.
Find an x402 tool →
Search public paid-agent endpoint metadata by task, recent source activity, Base network, and a hard USDC budget before preflight.
Run the verifier →
Free report for HTTP 402, x402 manifest, OpenAPI, catalog metadata, and fix recommendations.
Browse /discover →
Verification-backed JSON catalog. Use grade=ready or min_score=85 for strict agent payment.
Try the sandbox →
Run can-pay against simulated x402 allow, review, and deny fixtures before using real funds.
Agent payment firewall →
Turn readiness evidence into allow, review, or deny decisions before agents sign payment payloads.
Generate spend policy →
Create a starter JSON policy for budgets, review thresholds, deny rules, and Ontario can-pay calls.
Generate Agent Buyer Card →
State the exact job, inputs, output, price, preflight, evidence, and stop conditions an agent needs before selecting a paid tool.
Generate launch kit →
Create listing JSON, x402 and MCP drafts, an official Registry server.json draft, CI workflow, README copy, and registry text.
Scan repository →
Catch implementation, HTTP 402 test, metadata, CI, documentation, and credential-filename gaps before deployment.
Check MCP Registry readiness →
Inspect a public registry record and live MCP metadata before agents select or pay for a tool.
Audit buyer discovery →
Test one paid endpoint across five live Bazaar buyer searches, observed ranks, quality fields, prices, and metadata gaps.
Score catalog readiness →
Combine live x402, GitHub, and MCP evidence into one provider launch score before asking agents to discover or pay.
Audit settlement authenticity →
Detect self-payments, classify disclosed related-party receipts, and measure payer concentration before calling x402 settlement count customer demand. Includes July 2026 adoption evidence context.
Price an x402 endpoint →
Model cost, retries, fees, margin, and agent budget before publishing a per-call amount that agents can parse.
Danish AI/MCP audit →
Fixed-scope trust audits for Danish teams using AI assistants, MCP tools, or agent payment workflows.
Pay per query →
Settlement-backed /api/agent/pay endpoint for paid policy answers and can-pay data.
Inspect reports →
Crawlable readiness evidence: score, warnings, remediation, history, and certification status.
Read the proof →
Implementation mistakes, manifest repairs, live benchmarks, and MCP paid-tool security checks.