ActivityWatch Analysis Skill

A Claude Code skill for productivity analysis using ActivityWatch data. Calculates focus scores, detects "death loops" (repetitive app switching), and generates actionable insights.

Features

• Calibration Mode: First-run setup that shows uncategorized apps, detects missing watchers, and guides personalization
• 80+ App Defaults: JetBrains IDEs, terminals, browsers, design tools, CRMs — works out of the box for most developers
• AI Agent Detection: Recognizes Claude Code, Codex, Aider, GitHub Copilot as productive workflows
• Dual Scoring: Productivity score (what you worked on) + Focus score (attention quality)
• Deep Browser Analysis: Site-level breakdown with productivity ratios (requires aw-watcher-web)
• Death Loop Detection: Identifies repetitive app switching patterns with fix suggestions
• Telegram Chat Categorization: Separate work chats from personal messaging
• Customizable Categories: JSON config to tune for your workflow

Requirements

• Python 3.8+
• ActivityWatch installed and running
• Recommended: pip install aw-client for direct API access
• Recommended: aw-watcher-web Chrome extension for site-level browser tracking

Install as Claude Code Skill

cp -r activitywatch-analysis-skill ~/.claude/skills/activitywatch-analysis

Quick Start

Step 1: Calibrate (first run)

# Show what needs configuring
python scripts/analyze_aw.py --fetch --from week --calibrate --config scripts/category_config.json

This shows:

• Uncategorized apps with significant usage time
• Whether aw-watcher-web is installed (critical for browser analysis)
• Telegram usage that may need chat-level categorization

Then edit scripts/category_config.json to classify your uncategorized apps.

Step 2: Analyze

# Today
python scripts/analyze_aw.py --fetch --from today --report

# This week
python scripts/analyze_aw.py --fetch --from week --report

# Specific dates
python scripts/analyze_aw.py --fetch --from 2026-04-01 --to 2026-04-05 --report

# With your custom config
python scripts/analyze_aw.py --fetch --from today --report --config scripts/category_config.json

# JSON output (for automation)
python scripts/analyze_aw.py --fetch --from today --config scripts/category_config.json

Date formats: today, yesterday, week, 7d, 2w, YYYY-MM-DD

Step 3: CSV Fallback

If you don't have aw-client:

1. Open ActivityWatch (http://localhost:5600) -> Raw Data -> Export -> CSV
2. python scripts/analyze_aw.py export.csv --report

Understanding Your Scores

Combined Score (0-100)

Range	Interpretation
80-100	Excellent - Deep work patterns, minimal distractions
60-79	Good - Solid productivity with room to improve
40-59	Moderate - Attention fragmented, review death loops
0-39	Needs work - High distraction, consider app blockers

Productivity vs Focus

• Productivity Score: Measures what you spent time on (deep work vs. entertainment)
• Focus Score: Measures how you worked (sustained attention vs. constant switching)

High productivity + low focus = good work but constantly interrupted.

Death Loops

Repetitive A-B app switches that fragment your attention.

Verdict	Meaning	Action
ai_assisted	AI coding agent active	Productive workflow
productive	Normal workflow (IDE - Terminal)	Consider split screen
mixed	Could go either way	Batch these activities
distracting	Attention leak	Block during focus hours

See references/blocking_guides.md for blocking tool setup guides.

Customizing Categories

Edit scripts/category_config.json:

{
  "my_product": {
    "weight": 0.8,
    "description": "My SaaS product work",
    "apps": ["MyApp"],
    "titles": ["myapp.com", "MyApp Dashboard"]
  }
}

Weight scale: 1.0 deep work, 0.7-0.9 productive, 0.5 mixed, 0.3 shallow, 0.0 neutral, -0.2 to -0.5 distracting

Telegram Chat Categorization

Separate productive Telegram chats from personal messaging:

{
  "telegram_chats": {
    "product_work": {
      "weight": 0.8,
      "patterns": ["Team Chat", "Project Alerts", "bot-monitoring"]
    },
    "content_creation": {
      "weight": 0.7,
      "patterns": ["My Channel", "Content Ideas"]
    }
  }
}

Unmatched chats default to communication_personal (weight 0.1).

Recommended Watchers

Watcher	What it tracks	Install
aw-watcher-window	Active app + window title	Included with ActivityWatch
aw-watcher-afk	Idle/active state	Included with ActivityWatch
aw-watcher-web	Browser tab URLs + titles	Chrome extension / Safari build
aw-watcher-vscode	VS Code file-level tracking	VS Code marketplace

Without aw-watcher-web, all browser time appears as one block ("Chrome") with no site-level breakdown.

Integration with Claude Code

As a Claude Code skill:

/activitywatch calibrate     — first-run setup, personalize categories
/activitywatch               — analyze today
/activitywatch week          — analyze this week
/activitywatch review        — coaching-style comparison to previous weeks

Files

activitywatch-analysis-skill/
├── SKILL.md                      # Claude Code skill definition
├── CLAUDE.md                     # Development guidance
├── README.md                     # This file
├── scripts/
│   ├── analyze_aw.py             # Main analyzer
│   ├── focus_guard.py            # App blocker for focus sessions
│   ├── category_config.json      # Customizable categories (edit this!)
│   └── focus_config.json         # Focus Guard configuration
└── references/
    ├── analysis_prompts.md       # Prompts for deeper analysis
    └── blocking_guides.md        # App blocking setup guides

Privacy

All analysis runs locally. No data leaves your machine unless you choose to share it.

License

MIT

Author

Bayram Annakov (@BayramAnnakov)