Payment risk · Self-reported

AgentGuard® Score

Check payment processing risk for merchants and autonomous AI agents. Get a risk score, how likely a payment processor is to approve you, payment providers that fit, fraud and compliance flags, and what to fix first.

Free. No account.

What comes back

Two self-reported checks. Each returns the fields below, filled from the answers you give. There is no score until you answer.

Payment risk checkPOST /api/payment-risk
score
0 to 100. Higher is lower risk.
band
the risk band the score falls in
approval likelihood
high, moderate or low
typical reserve
the hold a processor may ask for at this score
dispute ratio
your chargebacks against the Visa monitoring threshold
providers
payment providers that fit your industry and region
flags
fraud and compliance flags, with a fix where there is one
what to fix first
up to six steps, in order
AI agent scorePOST /api/score
score
0 to 100, from five answers
tier
the band the score falls in
what decided it
each answer, and what was not verified
what to do next
the controls you reported missing, and where to set each one

Both scores cover what you report and verify nothing. Verified enforcement is a different check, made from a signed bundle further down.

Payment risk check

For a business that takes card payments, or an AI agent that sells on its behalf. Answer from the last 30 days. The first four questions are required.

Fraud tools you use

Leave all unchecked if you use none.

Documents you have ready

Leave all unchecked if none are ready.

Your answers stay on this page until you press Check payment risk. Then the answers above, and only those, go to this site (POST /api/payment-risk), which checks them against the published options and passes them to the hosted scoring service for the result. We ask for no name or email and do not store the answers; one count per completed check (the risk band and the day) is kept. The privacy policy has the detail.

Score an AI agent that moves money

Five questions about the controls around an agent that pays or gets paid: who is accountable, where its money sits, its limits and its records. The score covers what you report. It verifies nothing.

Is a named human accountable for this agent, and can a third party verify who that is?
Are per-transaction and daily limits enforced somewhere the agent cannot change them?
Is every payment action logged with a timestamp the agent cannot edit?

Your answers stay on this page until you press Score this agent. Then the five answers go to this site (POST /api/score), which passes them to the hosted scoring service to compute the score, with no email and without asking for a report link. One count per completed score (that it ran from this page, and the day) is kept.

Verified · Not a score

Check a signed bundle

A signed activity bundle from the AgentGuard plugin is the enforcement record itself, not a questionnaire. The signature check runs in your browser. Nothing is sent until you click Publish.

Publishing makes the complete bundle public, including its label if supplied, limits, counts and public key. Review it first. Save the private deletion link returned at publication.

Planner · Not a score

Plan your agent's rules

Choose the work and the decisions you want to keep. Get rules to review and a plain picture of your AI bill. This planner runs in your browser and sends nothing.

What should your agent do?

Choose all that apply.

What must it never do without asking you?

Choose all that apply. The result separates plugin checks from settings you need in your own tools.

This planner runs in your browser. Nothing you choose here is sent.

The bill above is a planning estimate. What a session actually spent comes from Burn, which reads the sessions saved on your computer and sends nothing:

npx agentguard-burn why

API for developers and agents

Both checks are plain HTTPS endpoints on agentguard.run. No key. JSON in, JSON out. Requests are rate limited per address.

Payment risk check

Send industry, monthly_transactions, total_disputes and fraud_disputes. Optional: region, business_age, avg_transaction_value, refund_rate, customer_mfa, checkout_platform, and the lists fraud_tools, current_psp and compliance_readiness. Values use the option ids from the form above. Any other field is refused.

curl -s https://agentguard.run/api/payment-risk \
  -H 'Content-Type: application/json' \
  -d '{"answers":{"industry":"saas","monthly_transactions":"500-2000","total_disputes":"1-5","fraud_disputes":"0","region":"US","fraud_tools":["3ds","cvv"]}}'

The reply carries score (0 to 100), band_label, approval_likelihood (high, moderate or low), reserve_estimate, recommended_providers, flags, recommendations, the dispute ratio in vamp, and a disclaimer.

AgentGuard Score for an AI agent

curl -s https://agentguard.run/api/score \
  -H 'Content-Type: application/json' \
  -d '{"answers":{"agent_type":"Customer service agent","human_sponsor":true,"wallet_type":"None yet","transaction_limits":true,"audit_trail":true}}'

Every one of the five answers is required. The reply carries score, tier, a category breakdown and factors with recommendations.

Inside Claude Code and Codex

The AgentGuard plugin includes the agentguard-score skill and the agent_score MCP tool. Your agent asks the five questions, names the service it will send them to, and sends them only after you say yes.

Results are estimates from the answers you give. They are not a guarantee that any payment processor will approve you, and they are not legal or compliance advice.

AgentGuard does not apply to payment providers for you and does not see your processor account. Provider names are a starting list for your industry and region.