Starting this weekend, I built HAR Insight a browser-based tool for looking at HTTP Archive (HAR) files. You can try it at har.thelazysre.com

The main idea: analysis runs on your machine. Your data doesn’t go anywhere unless you explicitly use the optional sharing or AI features. I wanted something I could use (and point others at) without uploading sensitive captures to a third party. Here’s what it does and why I built it that way.

The Problem I Wanted to Solve

When you need to analyze network traffic, you usually have two options. Both have downsides.

- Cloud-based analysers: You upload your HAR file to some web service. HAR files often contain cookies, API tokens, or other sensitive data. Many teams aren’t allowed to send that to external sites and even when they are, it’s not always something you want to do.

- Browser DevTools: Chrome and Firefox can record and inspect network activity. They’re powerful but tied to a single session. Sharing a clear view with someone else, or comparing two captures side by side, is awkward. The UI is also very developer-focused.

I wanted something in the middle: serious analysis in the browser, with optional sharing when you need it. That’s what HAR Insight is.

What HAR Files Are (Quick Refresher)

HAR stands for HTTP Archive. It’s a standard JSON format that describes HTTP requests and responses. Browsers can export their network activity as a HAR file URLs, headers, timing, and optionally response bodies.

People use them to debug slow pages, investigate errors, or audit security. Support teams often ask customers for a HAR to reproduce an issue.

What I Put in It

Privacy by default. Everything runs in your browser. Parsing, filtering, stats, waterfall it's all local. Nothing hits a server unless you use the optional “share” or “AI insights” features.

That was non-negotiable for me: I wanted to be able to point colleagues at the app and say “drop your HAR here” without worrying where the file goes.

Enough for real work. You get a proper waterfall (and Gantt-style) with DNS, connection, wait, and download phases. You can filter by URL (including regex), method, status code, or domain, and search inside response bodies. I added panels for performance (slow requests, large payloads, uncompressed stuff), security (missing headers, insecure cookies, mixed content), cache behavior, third-party breakdown, and content types. You can rearrange the panels and the layout is remembered. Multiple HAR files in tabs, and a side-by-side comparison of two captures. Export to HAR or CSV. Copy any request as cURL. That covers what I need when debugging or reviewing a capture.

Optional sharing and AI. When you do need to share a capture, you can get a short-lived link (backend stores it temporarily, then deletes it no account). There’s also an optional AI pass for performance and security suggestions. Both go through a Cloudflare Worker; the rest of the app works offline and doesn’t talk to any server.

No sign-up. You open the site, drop a file, and you’re in. I kept it that way on purpose; nice for ad-hoc use and for sending to non-technical people who just need to export a HAR and open a link.

Who I Had in Mind

If you debug slow or broken requests, ask customers for HARs, or review security/headers, it should work well. Same if you just want to compare two captures or keep a few HARs open in tabs without uploading them anywhere. I’ve sent the link to support folks who aren’t developers and they’ve managed fine.

A Few Caveats

Very large HAR files (hundreds of MB or tens of thousands of requests) can make the browser sluggish typical sizes are fine. You can’t edit or redact a HAR inside the app; if you need to strip sensitive data before sharing, you do that elsewhere. AI and share links require the backend; the rest works fully client-side and even offline (it’s a PWA). You need a modern browser.

Try It

No install, no account.

Drop a HAR file (export one from Chrome DevTools → Network → right‑click → “Save all as HAR with content”), and poke around.

It’s planned to be open source on GitHub (just need to open the repo to the public); feedback and contributions will be welcome.

If you work with HAR files and have been wishing for something that doesn’t upload your data by default, I hope this helps.