Proxie

Proxie is your personal AI agent. Publish your content, and your Proxie handles inbound communication on your behalf.

Folder Structure

Monorepo using npm Workspaces:

• api/: Backend Node REST API server
• client/: Frontend React app
• data/: Dataset (currently static, deploy via /scripts/content.js)
• infra/: Infrastructure definition for Kubernetes deploy
• proxy/: NGINX config for ingress to client and api
• scripts/: Helper zsh scripts (GCP setup, deploy, content)

Prereqs

• Node 18+
• npm 9+
• Docker (for proxy and deploy)
• gcloud CLI (for GCP scripts)

Install

npm install

This installs workspace dependencies for client, api, and proxy.

Config

The client fetches from /api/*. Configure backend host via VITE_API_BASE_URL in client/.env.local.

Run (local dev)

• API standalone (port 3001):

  npm run dev:api

• Client standalone (port 5173 by default):

  npm run dev:client

• Client + API concurrently:

  npm run dev

VSCode Debug (Optional)

• API and Client launch configs live in .vscode/launch.json.

Reverse Proxy (Optional)

NGINX is used primarily for deployment to handle ingress for client/dist and /api to http://localhost:3001.

• Build client and start API:

  npm run build && npm run dev:api

• Start NGINX proxy (Docker):

  npm run proxy

Build & Deploy

Build and deploy is handled by ./scripts/deploy.zsh to create a Docker image and release to GCP as a knative services.yaml

Note: Sensitive config should use GCP Secret Manager. See .env.stage.yaml and .env.prod.yaml for examples.

GCP Setup

Need to run setup at least once:

./scripts/gcp-setup.zsh --project <PROJECT_ID>

Deploy Script

npm run deploy <stage|prod>