Personal APIs

Personal APIs

Data aggregation system for self-quantification and AI context enhancement

Architecture Overview

Built on Nuxt / Nitro with basic caching layer. All endpoints return JSON and support CORS for cross-origin requests from mobile apps and AI assistants.

Base URL: https://ejfox.com/api/

Core Philosophy

• AI Context Enhancement - Provide rich personal data to language models
• Self-Quantification - Track KPIs aligned with personal improvement goals
• Data Sovereignty - Own your data instead of just feeding the surveillance machine
• API-First Design - Choose services that offer data egress and avoid data silos

Active Endpoints

Health & Fitness

• /api/health - Apple Health aggregated metrics
• /api/health-raw - Raw health data for deeper analysis

Tracks: daily steps, exercise minutes, stand hours, distance, heart rate variability

Productivity & Learning

• mastodon-posts.ejfox.tools most recent posts from @[email protected]
• /api/rescuetime - RescueTime productivity analytics
• /api/leetcode - LeetCode problem-solving progress
• /api/monkeytype - MonkeyType typing speed metrics
• /api/chess - Chess.com rating and puzzle stats

Development

• /api/github - GitHub contribution patterns and repo activity
• /api/stats - Aggregated development metrics

Creative & Content

• /api/photos - Personal photo platform statistics
• /api/lastfm - Last.fm music listening history (10+ years)
• /api/words-this-month - Writing output tracking

Predictions & Analysis

• /api/predictions - Personal forecasting and goal tracking

Technical Implementation

Data Sources

• Apple Health - via HealthKit exports and processing
• GitHub API - commit patterns, repo activity, contribution graphs
• RescueTime API - time tracking and productivity metrics
• Chess.com API - game history and ratings
• Last.fm API - scrobbling data going back 10+ years
• LeetCode scraping - problem completion and difficulty tracking
• MonkeyType - typing test results and WPM progression

Caching Strategy

Most endpoints implement intelligent caching:

• Health data - Updates every 4 hours
• GitHub activity - Updates every hour
• Music listening - Updates every 30 minutes
• Static metrics - Cached for 24 hours

Response Format

All endpoints return JSON with consistent structure:

{
  "data": { /* endpoint-specific data */ },
  "meta": {
    "cached_at": "2025-05-28T10:30:00Z",
    "source": "api_name",
    "updated": "recent" | "stale"
  }
}

Same Data, Many Interfaces

The personal API enables multi-modal access - same underlying data, different interfaces for different contexts.

The Pattern

[Data Layer] (APIs aggregate from multiple sources)
     ↓
[API Layer] (consistent JSON, unified caching)
     ↓
[Interface Layer] (multiple, context-specific)
├─ AI assistants (Claude, Coach Artie get full context)
├─ Website widgets (live stats display)
├─ Mobile apps (health dashboard)
└─ CLI tools (terminal queries)

Data stays useless until something accesses it. The API makes personal data queryable by any interface.

Context-Driven Access

Context	Interface	Example
AI conversation	Claude/Coach Artie	"How much did I code today?" → API provides actual data
Morning review	Website dashboard	Visual stats, charts, trends
Quick check	Alfred/CLI	One-line query, instant answer
Deep analysis	Python/SQL	Full data export, custom analysis

Why This Matters

For AI context: Language models with access to actual behavioral data give more relevant, personalized responses. "You haven't committed in 3 days" beats generic advice.

For self-understanding: Same data viewed through different lenses reveals different patterns. The dashboard shows trends; the CLI answers specific questions.

For dogfooding: You are the user. The interface that's fastest for you is the one that matters.

Related

• Quantified Self - Philosophy of tracking
• Scrapbook-core - Another multi-interface system
• Coach Artie - AI that consumes these APIs
• Same Data Many Interfaces - The design pattern
• Building Tools for Yourself - Why personal tools work