Philosophy

Design Principles

Snakeway is built around a few ideas:

• Explicit over implicit: Nothing happens unless you configure it.
• Programmable, not declarative-only: Real logic requires real code.
• Safe by default: User-defined logic runs in constrained environments.
• Observable from day one: Traffic you can’t see is traffic you don’t control.

Where Snakeway Fits

Snakeway sits between the client and the web service, like nginx or any other reverse proxy.

This allows Snakeway to make decisions at the edge before traffic ever hits application code.

Common use cases include:

• Early rejection of invalid or abusive requests
• Rules engines and request classification
• Header normalization and enrichment
• Static file serving alongside proxied traffic
• Structured access logging and observability

The Core Extensibility Model

Everything in Snakeway revolves around devices.

A device is a unit of logic that runs at a specific point in the request lifecycle.

Devices can:

• Read or modify request headers and bodies
• Make routing decisions
• Short-circuit requests with a response
• Observe traffic for logging or metrics
• React to upstream responses or errors

Devices are executed in a strict, ordered pipeline.

This is not middleware in the traditional web-framework sense. Devices are plugins that hook into the underlying Pingora data plane’s HTTP pipeline at various points. Specific hooks are optional; which hooks are implemented depends on the device. For example, the Identity device only implements the on_request hook.