sanity-plugin-gemini-ai-images-serverless

Serverless adapter for the Sanity Gemini AI Image Generator plugin.

Works with Vercel Functions, Netlify Functions, AWS Lambda, and other serverless platforms.

Installation

npm install sanity-plugin-gemini-ai-images sanity-plugin-gemini-ai-images-serverless @google/genai

Setup

1. Add Plugin to Sanity Config

// sanity.config.ts
import {defineConfig} from 'sanity'
import {geminiAIImages} from 'sanity-plugin-gemini-ai-images'

export default defineConfig({
  plugins: [
    geminiAIImages({
      apiEndpoint: '/api/gemini/generate-image', // or your custom path
    }),
  ],
})

2. Create Serverless Function

Vercel Functions:

// api/gemini/generate-image.ts
import {handler} from 'sanity-plugin-gemini-ai-images-serverless'

export default handler

Netlify Functions:

// netlify/functions/generate-image.ts
import {handler} from 'sanity-plugin-gemini-ai-images-serverless'

export {handler}

AWS Lambda (with API Gateway):

// lambda/generate-image.ts
import {handler} from 'sanity-plugin-gemini-ai-images-serverless'

export {handler}

3. Set Environment Variables

Vercel:

# .env or Vercel Dashboard
GEMINI_API_KEY=your_api_key_here
SANITY_PROJECT_ID=your_project_id
SANITY_DATASET=production

Netlify:

# netlify.toml or Netlify Dashboard
[build.environment]
  GEMINI_API_KEY = "your_api_key_here"
  SANITY_PROJECT_ID = "your_project_id"
  SANITY_DATASET = "production"

AWS Lambda: Set via AWS Console → Lambda → Configuration → Environment variables:

• GEMINI_API_KEY: your_api_key_here
• SANITY_PROJECT_ID: your_project_id
• SANITY_DATASET: production

Get your Gemini API key from Google AI Studio.

Important:

• The Sanity project ID and dataset are required for authentication verification
• These should match your Sanity Studio configuration

Platform-Specific Notes

Vercel

• Function timeout: Default is 10s, but image generation may take longer
• Increase timeout in vercel.json:

{
  "functions": {
    "api/gemini/generate-image.ts": {
      "maxDuration": 60
    }
  }
}

Netlify

• Function timeout: Default is 10s for free tier, 26s for Pro
• For longer timeouts, upgrade to Pro or use background functions
• Configure in netlify.toml:

[functions]
  directory = "netlify/functions"

[[functions."generate-image"]]
  timeout = 26

AWS Lambda

• Default timeout is 3s - increase to at least 60s
• Configure memory to at least 1024MB for better performance
• Enable function URL or use API Gateway

Cloudflare Workers

Not currently supported due to Workers' runtime limitations. Consider using Cloudflare Pages Functions instead, which support Node.js APIs.

Custom Endpoint Path

If your function is at a different path, update the Sanity config:

geminiAIImages({
  apiEndpoint: '/api/custom/ai-images',
})

Requirements

• Node.js 20.0.0 or higher
• Sanity Studio v4
• Serverless platform that supports Node.js runtime

Authentication

This adapter automatically verifies that requests come from authenticated Sanity Studio users. The authentication flow works as follows:

1. The core plugin (in Sanity Studio) retrieves the user's token from the Sanity client
2. The token is sent in the Authorization: Bearer <token> header with each request
3. The adapter verifies the token against the Sanity API before processing the image generation request

This ensures that only authenticated Sanity Studio users can generate images through your API endpoint.

Troubleshooting

Function Timeout

Image generation can take 10-30 seconds. Increase your function timeout:

• Vercel: Use vercel.json (see above)
• Netlify: Upgrade to Pro for 26s timeout
• AWS Lambda: Increase in function configuration

CORS Errors

The handler includes CORS headers by default. If you still see errors, check your platform's CORS configuration.

Cold Starts

First request may be slow due to cold starts. This is normal for serverless functions.

Unauthorized Error (401)

If you receive "Unauthorized" errors:

• Ensure you're logged into Sanity Studio
• Verify SANITY_PROJECT_ID and SANITY_DATASET environment variables are set correctly
• Check that the environment variables match your Sanity Studio configuration
• Make sure your serverless platform has the environment variables configured
• Note: Use SANITY_PROJECT_ID (not NEXT_PUBLIC_SANITY_PROJECT_ID) in serverless environments

License

MIT © Nick Jensen