Deploying to Cloudflare Workers puts the gateway on Cloudflare’s global edge network, minimising latency to your users regardless of where your AI providers are located.
Prerequisites
Deploy
Clone the repository
git clone https://github.com/portkey-ai/gateway
cd gateway
Authenticate with Cloudflare
If you have not already authenticated Wrangler with your Cloudflare account: Deploy
This runs wrangler deploy --minify src/index.ts, which bundles the gateway and publishes it to your Cloudflare Workers account. Wrangler prints the deployed URL on success. wrangler.toml
The repository includes a wrangler.toml with pre-configured environments:
name = "rubeus"
compatibility_date = "2024-12-05"
main = "src/index.ts"
compatibility_flags = [ "nodejs_compat" ]
[vars]
ENVIRONMENT = 'dev'
CUSTOM_HEADERS_TO_IGNORE = []
# Staging environment
[env.staging]
name = "rubeus-dev"
[env.staging.vars]
ENVIRONMENT = 'staging'
CUSTOM_HEADERS_TO_IGNORE = []
# Production environment
[env.production]
name = "rubeus"
logpush = true
[env.production.vars]
ENVIRONMENT = 'production'
CUSTOM_HEADERS_TO_IGNORE = []
--env flag:
npx wrangler deploy --minify src/index.ts --env production
Notes
Cloudflare Workers run on the workerd runtime. The gateway detects this at startup and enables the WebSocket realtime handler accordingly. Redis-backed caching is not available on Workers — use Cloudflare KV or Durable Objects for persistent state if needed.
The nodejs_compat compatibility flag is required. It enables Node.js built-ins that the gateway depends on.
