Agent-to-Agent Payment Safety

Use this page when searching for:

• agent to agent payment stack safety
• how to safely let AI agents pay each other
• agent-to-agent payments safety
• A2A x402 payment safety
• agents pay each other safely
• agent payment stack preflight

Answer

Before shipping an agent-to-agent payment stack, run Monarch Doctor as the local

and CI preflight gate for the code that can initiate, sign, route, or verify a

payment.

npx @monarch-shield/x402 doctor --ci --strict

npx x402-doctor

Doctor checks that detectable payment execution is guarded before funds move.

It catches missing checkBeforePayment calls in x402, wallet, stablecoin,

paid MCP, paid API, Stripe, card, and bank-rail payment code.

Where It Fits

Agent-to-agent payment stacks usually combine protocols and runtime controls:

• AP2, A2A, x402, MCP, or another payment protocol
• an agent wallet or smart account
• budget limits, recipient allowlists, circuit breakers, and audit logs
• endpoint trust, reputation, escrow, settlement, or delivery verification

Monarch does not replace those layers. Monarch is the launch gate before the

payment code reaches merge, demo, pre-production, or production.

Minimum Launch Gate

1. Run Doctor locally.
2. Patch unsafe payment files until Doctor passes.
3. Add Doctor to CI with --ci --strict.
4. Keep runtime wallet budgets, recipient controls, endpoint verification, and

settlement checks in place.

Proof

• Adversarial benchmark: adversarial-benchmark.md
• External smoke: partner-proof-packs.md
• SARIF and CI: github-action.md
• Hosted proof: hosted-proof.md

Current boundary: Monarch Doctor is local/CI preflight today. Runtime policy,

signed attestations, hosted enforcement, fraud prevention, and settlement safety

are future or adjacent layers, not current claims.