EJ Fox's Website

Personal website + blog: Nuxt 3 • Vue 3 • Tailwind • Docker

Quick Start

yarn install && yarn blog:import && yarn blog:process && yarn dev

Commands

Command	Description
yarn dev	Dev server (port 3006)
yarn build	Production build
yarn blog:import	Import from Obsidian
yarn blog:process	Process MD → JSON
yarn predict	Create cryptographic prediction
yarn webmentions	Send webmentions (last 7 days)
yarn webmentions --all	Send all webmentions

Features

Content Pipeline

Obsidian → Markdown → Unified/Rehype → JSON → Vue

content/blog/YYYY/*.md  →  content/processed/YYYY/*.json

Frontmatter

---
title: Post Title
date: 2024-01-01
dek: One-sentence description
tags: [tag1, tag2]
draft: false # Hide from lists
hidden: false # Skip processing
replyTo: https://example.com/post # IndieWeb reply (single or array)
---

IndieWeb

Feature	Details
Reply posts	replyTo: fetches OG data, shows context, sends webmention
Webmentions	Display likes/reposts/replies, moderation via data/webmention-moderation.json
Microformats	h-entry, h-card, u-in-reply-to on all posts
WebSub	Hub at pubsubhubbub.superfeedr.com

Predictions System

yarn predict --statement "X will happen" --confidence 75 --deadline 2025-12-31

• SHA-256 + Git commit + optional PGP signing
• AI quality analysis via OpenRouter
• See docs/PREDICTIONS.md

Stats Aggregation

Two endpoints for personal metrics:

• /api/stats - Full stats with arrays and nested data (~2.4KB)
• /api/stats-lite - Lightweight for iOS Shortcuts (~345 bytes, 86% smaller)

Aggregates: GitHub, Chess.com, LastFM, RescueTime, Letterboxd, Discogs, Kalshi

See docs/STATS-API.md and docs/IOS-SHORTCUTS-EXAMPLES.md for usage

Gear Inventory

CSV-based gear tracking at /gear with weight calculations, Tuftian visualizations

Sidenotes

Footnotes auto-convert to margin notes on desktop (Tufte-style)

Deployment

Push to main → GitHub Action SSHs to VPS → pulls, builds, restarts Docker

Manual: ssh vps "cd /data2/website2 && git pull && yarn build && docker-compose up -d --build"

Environment

CLOUDINARY_CLOUD_NAME=
CLOUDINARY_API_KEY=
CLOUDINARY_API_SECRET=
YOUTUBE_API_KEY=
YOUTUBE_CHANNEL_ID=

Directory Structure

content/blog/       # Source markdown by year
content/processed/  # Output JSON
server/api/         # API endpoints
components/         # Vue components
scripts/            # Processing scripts
data/               # Config files (moderation, gear CSV)

Debug

DEBUG=true yarn blog:process
CHECK_LINKS=true yarn blog:process  # Link health check
AUTO_FIX_LINKS=true yarn blog:process  # Auto-fix with archive.org

TODO

• Rerun node scripts/hydrate-cloudinary-cache.mjs until it reports 0 missing metadata (it resumes from cache).
• Investigate remaining x failures in the hydrator output (missing assets vs permissions).
• Document the hydrator runtime/usage and add it to the Commands table once stable.
• Rerun the hydrator to reduce the remaining ~331 missing entries (current baseline).
• Clean malformed Cloudinary URLs in markdown (e.g. ...png)![Screenshot) so they can be hydrated.

---

MIT License