🔒 All inputs are processed locally in your browser. No data is transmitted, stored, or logged. Paste only synthetic or anonymised AP2 mandates — never live customer data.
A visual guardrail for machine-readable payment agent mandates.
Paste an AP2 JSON mandate on the left. The tool parses every constraint, renders it as a shield-gated flow chart through the agent's decision pipeline, and compiles a validated mandate ready to hand off to your MCP agent runtime. Every shield is a contract: the agent must satisfy all of them before settling a payment.
AP2 Mandate · Principal
— —
mnd_—
Agent ID
—
Scope
—
Issued
—
Expires
—
Decision Pipeline · Guardrail Flow— shields
01
Inbound Payment Intent
user_signal · cart · context
02
Shielded Constraints
every shield must clear
03
Resolution Branch
deterministic outcomes
Outcome Tree
▲ ALLOW
All shields cleared, no escalation triggered.
→ Settle on preferred rail → Emit ap2.outcome.allow event
◆ REVIEW
Any REVIEW shield triggers, or amount in escalation band.
→ Hold payment, notify principal → Wait for human confirmation
■ DENY
Any DENY shield triggers, or ENFORCE contract violated.
→ Reject with reason code → Emit ap2.outcome.deny event
Compile to MCP-ready AP2 mandate
Generates a signed, version-stamped JSON envelope your MCP agent runtime can consume directly. Constraints are normalized, severity flags are resolved, and an agent_instructions block is appended in natural language so an LLM-backed agent can reason about edge cases.
Agentic Runtime
Building agentic payment infrastructure?
We design the deterministic AP2 / MCP policy layer that runtimes like this one depend on. If you're putting agents anywhere near money, let's pressure-test your mandate architecture.