Hey Hunters! 👋 I'm Adrian, founder of AnterisLab. I built Guard after seeing AI agents get full payments:execute scope with a self-reported agent_id string. One prompt-injection away from a double charge. So we flipped it: LLM proposes (untrusted) → deterministic PDP decides → human approves → PEP alone executes toward Stripe. Today you get: - Open-source TS SDK (MIT) + public sandbox, no signup - Propose/Approve/Execute with SoD + 5min single-use approval - Server-side idempotency, hash-chained audit, 4-level kill-switch - Warm-path PDP decision <15ms p95 — repro with npm run bench I'd love your brutal feedback: 1. Is 5min approval / 10min intent TTL sensible for your agent? 2. Stripe or Adyen first for you? I'll be here all day. Thanks! 🙏 Try: https://anterislab.com/?utm_sour... Repo: https://github.com/AnterisLab/an...

AnterisLab Guard

Secure payments for AI agents in <15ms

Secure payments for AI agents in <15ms