AI-Assisted Deployment

The codebase includes deployment context files that help AI agents configure and troubleshoot deployments. This guide explains how to work with AI on deployment tasks.

Pro Tip

AI agents read llms/commands/deploy.md for deployment workflows. Referencing this file helps AI set up deployments correctly.

Context Files for AI

`llms/commands/deploy.md`

Deployment workflow guide covering:

Pre-deployment validation (pnpm validate)
Platform-specific build configurations
Environment variable requirements
Post-deployment verification steps

When working with AI on deployment:

"Read llms/commands/deploy.md and help me deploy to Vercel"

Effective AI Prompts

Clear and Specific

Common AI Workflows

Initial Deployment Setup

1. "Read llms/commands/deploy.md"
2. "Configure Vercel deployment for apps/marketing"
3. "List all required environment variables"
4. "Create deployment checklist"

AI will generate platform-specific configuration and environment variable lists.

Troubleshooting Build Failures

"The Vercel build is failing with 'Module not found: @workspace/ui'.
Read the build logs and suggest fixes."

AI will analyze error messages and suggest configuration changes.

Environment Variable Setup

"List all required environment variables for production deployment.
Include BetterAuth, Convex, and UseAutumn configuration."

Validation After AI Configuration

Verify Build Command

Check that AI used correct monorepo build command:

# Should be:
cd ../.. && pnpm turbo run build --filter=marketing

# Not:
pnpm build

Check Root Directory

Verify root directory points to correct app:

For marketing: apps/marketing

For dashboard: apps/dashboard

Validate Environment Variables

Ensure all required variables are set: APP_BASE_URL, DASHBOARD_BASE_URL, NEXT_PUBLIC_CONVEX_URL, BETTER_AUTH_SECRET, GOOGLE_CLIENT_ID, GOOGLE_CLIENT_SECRET, AUTUMN_SECRET_KEY.

Test Locally

Run validation before deploying:

pnpm validate

Common AI Mistakes

Best Practices with AI

1. Reference Deployment Context

"Read llms/commands/deploy.md and..."

Ensures AI follows established deployment patterns.

2. Specify Platform

"Deploy to Vercel..." or "Configure for AWS Amplify..."

Be explicit about target platform.

3. Request Checklist

"Create a deployment checklist for production"

AI will generate step-by-step verification list.

4. Validate Configuration

Always review AI-generated configuration before deploying. Check:

Build commands use Turbo with --filter
Root directory points to correct app
Environment variables include all required values
OAuth callbacks updated for production domain

Deployment Checklist for AI

When AI configures deployment, verify:

Build command uses monorepo Turbo pattern
Root directory set to apps/marketing or apps/dashboard
Install command includes --prod=false
All required environment variables configured
Convex environment variables synced
OAuth provider callbacks updated
pnpm validate passes locally
Post-deployment health checks planned

Templates for AI

Point AI to these patterns:

Vercel Deployment

Git-integrated deployment with automatic previews

AWS Deployment

Amplify, ECS, or Lambda deployment options

Netlify Deployment

Edge functions and form handling

Release Checklist

Pre-deployment security verification

Overview - Deployment strategy overview
Release Checklist - Security verification
Vercel - Vercel deployment guide
AWS - AWS deployment options

AI-Assisted Deployment

Good prompts

Vague prompts (avoid)

Wrong build command

Missing --prod=false flag

Forgetting Convex environment sync

Not updating OAuth callbacks

Vercel Deployment

AWS Deployment

Netlify Deployment

Release Checklist

On this page