Hey Product Hunt 👋 I kept hitting the same wall while building local AI agents: the moment my agent needed to react to something happening *out there* — a GitHub push, a Linear issue, a Stripe payment — it couldn't. Webhooks need a public URL. My agent runs on a laptop behind NAT. Every workaround meant opening ports, running ngrok, or writing my own receiver just to verify signatures and queue payloads. EventPort is the missing inbox. Create a subscription, get one public URL, point any platform at it. Your agent polls `GET /events` on its own schedule and consumes events on read. Nothing ever connects inbound to your machine — no ports, no tunnels. What's inside: • Signature verification at the edge for GitHub, Linear, Stripe, Shopify, Slack, HubSpot, Calendly and Typeform. Anything else → custom, with your own signing secret. • Two tokens per subscription, with separate roles: `egw_*` can only post webhooks, `egc_*` can only poll. A leaked consumer token can't inject events. • A durable queue on Cloudflare D1. If your agent is offline, events wait — adapters also dead-letter locally so nothing is silently lost. • Three install guides an agent can follow by itself: /agent/dsh, /agent/pi and /agent/exec (the last one works with any CLI agent). It's MIT licensed and self-hostable: one Cloudflare Worker + D1, plus the Next.js dashboard. Deploy your own in minutes, or just use the hosted version. One thing I'd genuinely like your help with: which source should I build next — RSS, webpage monitoring, or X/YouTube? The roadmap is deliberately short right now, and I'd rather build what you actually need. Happy to answer anything about the architecture, or about the at-most-once delivery tradeoff. — tudou527

EventPort

A public webhook inbox for local AI agents

A public webhook inbox for local AI agents