veryCrunchy Site

Nuxt 3 frontend with a Directus-powered content layer for:

• site settings
• blog posts
• photo archive

Setup

Make sure to install the dependencies:

pnpm install

Create a .env file with the Directus connection details:

NUXT_PUBLIC_DIRECTUS_URL=https://your-directus-instance.example.com
DIRECTUS_TOKEN=your_directus_static_token

DIRECTUS_TOKEN should be a read token with access to the public-facing collections below. If your Directus instance exposes these collections publicly, you can omit the token and the site will still render.

Directus Collections

The site expects these collections:

site_settings

Use a single item with these fields:

• site_name string
• site_tagline text
• site_description text
• github_url string
• support_url string
• nav_cta_label string
• nav_cta_url string
• posts_label string
• photos_label string

posts

• status string
• slug string
• title string
• excerpt text
• content text or rich text HTML
• cover_image file
• published_at datetime
• featured boolean
• tags JSON string array

photos

• status string
• slug string
• title string
• description text
• image file
• published_at datetime
• taken_at datetime
• location string
• camera string
• lens string
• featured boolean
• tags JSON string array

Only items with status = published are exposed on the public site.

Development Server

Start the development server on http://localhost:3000:

pnpm run dev

Production

Build the application for production:

pnpm run build

Locally preview production build:

pnpm run preview

Agent Photo Ingest

There is a reusable photo ingest CLI for agent-driven uploads:

pnpm run photos:prepare -- --manifest /tmp/photo-manifest.json /absolute/path/to/photo-1.jpg /absolute/path/to/photo-2.jpg
pnpm run photos:validate -- --manifest /tmp/photo-manifest.json
pnpm run photos:upload -- --manifest /tmp/photo-manifest.json

Optional Directus schema helpers:

pnpm run photos:setup-motion
pnpm run photos:setup-shots
pnpm run photos:setup-taxonomies
pnpm run photos:setup-timelines

The intended workflow is:

1. Run photos:prepare with exported image paths or --paths-file.
2. Let the agent inspect the images and ask a couple of concise questions about the journey, story, and grouping.
3. Fill in each photo's title, description, slug, and setSlugs in the generated manifest.
4. Optionally add sets definitions for custom photoset titles and descriptions.
5. Run photos:validate, then photos:upload.

The manifest includes EXIF-derived metadata, suggested auto-set candidates, and a storyContext section the agent can use while writing titles and descriptions.

TIFF uploads are converted to JPEG automatically before being sent to Directus.

If an edited export is missing GPS or place fields, the ingest CLI will also look for matching original raw and XMP sidecar files under PHOTO_INGEST_ORIGINAL_ROOTS (defaults to /mnt/h/DCIM) and merge that metadata into the manifest and converted upload.

If GPS exists but no readable place name does, photos:prepare will reverse geocode the coordinates through Nominatim and cache the result locally. You can override this with PHOTO_INGEST_REVERSE_GEOCODE_URL, PHOTO_INGEST_REVERSE_GEOCODE_CACHE, PHOTO_INGEST_REVERSE_GEOCODE_USER_AGENT, or disable it with PHOTO_INGEST_REVERSE_GEOCODE=0.

If a photo should behave like a motion sequence, add motionFrameSourcePaths to that photo entry in the manifest. Those files will also be converted to JPEG when needed and linked as motion frames before the final hero image.

If a photo needs attached still variants under the same slug, add a shots array to that photo entry. Each shot can include sourcePath, role, title, and description. This is intended for alternates, details, and nearby frames that belong to one published photo without turning into a full set.

You can also add structured metadata and story placement:

• locationMeta for structured places
• cameraMeta and lensMeta for lookup-backed gear metadata
• timelineEntries for placing a photo inside one or more story timelines with optional chapterTitle and storyText