Follow the request.
One stable endpoint for your GitHub webhooks and OAuth callbacks—even when your development environments keep changing.
Open source. Self-host today. Mildly conspiratorial.
01 / The usual suspects
For local development, shared repos,
and sandboxes with a short life expectancy.
Rebuild a preview or move to a different dev machine. Register its current address with Pepe; GitHub and Google keep calling the same public endpoint.
The URL changed. The trail didn’t.
Connect your environment ↗Everyone has a branch and their own environment. Pepe remembers which workspace owns each PR, so its comments and reviews reach the right developer.
No “whose laptop got that?”
Follow the PR ↗Give each sandbox its own Pepe session and webhook queue. Your agents can work in parallel, with each PR’s events routed to its registered workspace.
Short-lived hosts. Long-term memory.
Wire it into your agent ↗If a webhook receiver goes offline, Pepe holds its events. Reconnect to pick them up, or replay a stored delivery when you need another look.
The webhook can wait.
See queues and replay ↗02 / The theory
GitHub needs somewhere to call.
You have more than one somewhere.
Give your webhooks and OAuth callbacks one stable place to land.
The meeting pointPoint GitHub at Pepe. One front door, however many hosts sit behind it.
The incoming evidencePepe routes the request to the host that should handle it.
It was connected all along03 / The case file
A local environment. A preview.
The real thing. Same front door.
An illustrative routing diagram.
No actual requests are sent.
See? It’s all connected.
Case status: open
The open-source router is available today. Self-host it now, or join the private beta if you’d rather we host it.