Posts

  • A Spending Guardrail for Agentic Payments — Hands-On x402

    ai-agentsagentic-paymentsx402tutorial

    Let Claude pay an x402 testnet paywall without holding the wallet key, then test payment limits, rolling budgets, and duplicate protection.

  • Your Coding Agent Will Always Tell You It's Safe

    ai-agentssecurityagentic-securityagent-verifiability

    Agentic security starts from agent trust, and trust starts from verifiability: an agent can't monitor itself or declare itself secure. Why defense in depth needs a watcher underneath — read-only, open source, local-only — and why I built perch.

subscribe via RSS