SpendVeto stops AI agents from overspending on x402 and MCP payments. Policy caps, human approvals, nested budgets, and a kill switch enforced before any payment settles. Open source, Apache-2.0, 280 assertions.
I've been in crypto since I was 17, and I've watched what automated systems do to a wallet when nothing's standing in the way. Now AI agents can hold wallets and pay for things autonomously — and almost nobody's building the layer that stops one from overspending.
So I built SpendVeto: policy caps, human approval above a threshold, nested budget delegation, a kill switch — all enforced before a payment settles, not after. It's open source, Apache-2.0, and every claim on the site is backed by a real, running test (280 of them).
No users yet, no revenue — I'm building this in public and looking for design partners running real agent fleets who'd want to try it. Would genuinely love feedback from anyone here thinking about the buyer side of agentic commerce.
If anyone wants to contribute to the project — code, docs, ideas, whatever — DM me here on https://x.com/RevanthRajeev . It's a solo build right now and there's a lot of surface area here.