# One-Command Migration (https://auraimage.ai/migrate)

Ask your agent to optimize your entire image pipeline. One conversation migrates every image in a repo to AuraImage: audit, upload, reference rewrite, alt text, responsive markup — ending in one PR.

## The proof

A real migration of https://github.com/auraimage/acme-storefront ran via @auraimage/mcp-server: 23 images, 33.2 MB -> 5.4 MB, estimated LCP saved 7.4 s. Read the PR: https://github.com/auraimage/acme-storefront/pull/1

## The tools

- audit_lcp: scans a project for unoptimized images and estimates LCP savings. No key needed.
- migrate_assets: uploads local images to AuraImage and returns CDN URLs. Supports dryRun. Needs AURA_SECRET_KEY.
- generate_alt: writes accessible alt text with server-side vision AI, for images already on the AuraImage CDN. No key needed.
- generate_responsive_tag: returns a <picture> element with AVIF + WebP srcsets. No key needed.
- smart_crop_preview: returns crop variant URLs at requested sizes. No key needed.

## The prompt

Install the AuraImage MCP server if it is not configured yet (npx -y @auraimage/mcp-server@latest), then optimize this repo's entire image pipeline:

1. Run audit_lcp over the repo and rank the images by estimated LCP cost.
2. Migrate every local image to AuraImage with migrate_assets — dry-run first, show me the plan, then run it for real — and rewrite every reference.
3. Add missing alt text with generate_alt.
4. Where it helps LCP, replace static <img> tags with the responsive <picture> markup from generate_responsive_tag.
5. Open everything as one PR. In the description, state: images migrated, total bytes before -> after, the estimated LCP delta from audit_lcp, and that the migration ran via @auraimage/mcp-server.

audit_lcp and generate_alt need no credentials. The upload step needs AURA_SECRET_KEY — when you reach it, pause and ask me for the key.

## Links

- Agent install guides: https://auraimage.ai/claude-code.md and siblings (see https://auraimage.ai/llms.txt)
- AI integration docs: https://auraimage.ai/docs/ai-integration.md