Tare

Analyze any web page for unnecessary code, surveillance trackers, unused JavaScript & CSS, and asset bloat. Generates a self-contained HTML report and a machine-readable JSON report.

What it checks

• Tracker detection — matches all network requests against a curated database of 100+ known trackers across 7 categories: analytics, advertising, fingerprinting/session recording, social embeds, A/B testing tools, chat widgets, and tag managers
• Unused JavaScript — uses the Chrome Coverage API to measure what percentage of each script is actually executed on page load
• Unused CSS — measures how much of each stylesheet is actually applied
• Third-party requests — counts, sizes, and categorizes all cross-origin requests
• Asset bloat — detects oversized images, missing lazy-loading, and excessive web fonts
• Scoring — Page Health Score (performance & bloat) and Privacy Score (tracking & surveillance), each 0–100
• Recommendations — actionable fixes with privacy-respecting alternatives (Plausible, Umami, Fathom, GoAccess, OpenReplay, GrowthBook, etc.)

Quick start

npm install
npx playwright install chromium
npm start
# Open http://localhost:3000

Requirements

• Node.js 18+
• ~300 MB disk for Chromium (downloaded once by Playwright)

Output

Each analysis generates two downloadable files in reports/:

• web-bloat-report.html — self-contained, fully offline-readable report
• web-bloat-report.json — raw data for further processing

Reports are kept for 10 minutes then cleaned up automatically.

Privacy

This tool runs entirely on your own machine. No data is sent anywhere except to the URL you choose to analyze. The tracker database is embedded in the source — no external list fetching at runtime.

Persistence

Analyses are stored in Neon Postgres (DATABASE_URL, pooled). Schema auto-creates at boot. npm test needs no DB (uses pg-mem).

Architecture

server/
  index.js            Express server, routes, SSE progress streaming
  analyzer.js         Playwright orchestration (coverage, network, DOM)
  scoring.js          Page Health + Privacy scoring algorithms
  recommendations.js  Rule-based recommendation engine
  reportGenerator.js  HTML + JSON report generation
  jobs.js             In-process job state + EventEmitter registry
  data/
    trackers.js       Embedded tracker database (100+ entries)
public/
  index.html          Web UI
  style.css
  app.js              SSE client, result rendering
reports/              Generated report files (gitignored)

Open source — newsroom tech, by a journalist

This is one of OnlineJourno's fully open-source tools (MIT). OnlineJourno's products are fair-source; this one is a deliberate gift to the commons — no strings.

Why a journalist built this: most newsroom technology is built by engineers at news organisations, for news organisations. This isn't that. I'm a journalist — 25+ years in digital newsrooms — and I built this because readers' data leaks to ad-networks while a page crawls to load, and no one in editorial can see it. A page's privacy cost should be visible to the people who publish it, not buried in ad-ops. It's open because surveillance-free publishing shouldn't be a paid feature.

If you care about privacy, the open web, or journalism — this is an on-ramp. Contributions especially welcome on the tracker database — see CONTRIBUTING.md; start with an issue labelled good first issue.

Licence

MIT — fully open source, use it freely. See LICENSE.md.

Maturity: live. Solid: SSRF guard, 180-test suite incl. golden band/detector tests. Rough: openness audit missing from README (P5.13).

Suite licensing: Source-available under FSL-1.1; converts to Apache 2.0 after two years. The free tools (Tare, Crawl-Budget Analyser) are MIT. Pulse is a proprietary showcase of the engine.

Grading note: the reader-respect scale (A>=80, B>=65, C>=45, D>=25; Reader-Respecting through Egregious) is deliberately distinct from the OnlineJourno suite's A-F editorial bands.