Offline viewer with HAR Desk
This opens a small checkout capture from a made-up shop so you can see the whole inspector before using your own file. Drop a .har, choose one, or paste its JSON: the table, filters, request detail and summary work on the file in this tab, and nothing is uploaded.
- Drop your .har file on the page, choose it with the button, or paste the JSON. Files up to 50 MB are free.
- Read the summary: request count, failures, transferred bytes, the slowest and largest requests and the third-party hosts.
- Filter by status class, method, type, text, failed only or slow only, then click a row.
- In the detail pane open headers and cookies, query, request body, response body and the timing breakdown.
What to know
A HAR (HTTP Archive) file is a JSON log of what a browser did on the network: every request, its headers and cookies, the response, how long each phase took and, if the recording kept them, the bodies. Chrome, Edge and Firefox write them from the Network panel (Save all as HAR), and support teams ask for them because they show what the server really received and returned.
This viewer reads the file in your tab. No part of it is sent to us, which matters because a HAR usually carries session cookies and Authorization headers. It copes with the usual real-world gaps: missing fields, bodies stored as base64, several pages in one capture, and browser-specific fields that start with an underscore.
Two things are left out on purpose. WebSocket messages (the _webSocketMessages field some browsers add) are counted but not displayed, and binary or compressed bodies are described rather than dumped. The timing waterfall draws blocked, DNS, connect, send, waiting and receive; TLS time is already inside connect in the HAR format, so it is not drawn twice.
Updated · Cosmovex
Questions
Is my HAR file uploaded when I open it here?
No. The file is parsed in your browser tab. The only requests to our servers are the page itself, anonymous button-use events (never file content) and the Pro licence check if you have Pro.
Does it work for very large HAR files?
Free reads files up to 50 MB; Pro up to 250 MB. A browser holds the whole parsed file in memory, so a very large capture can be slow or fail on a small device. Recording fewer requests or one page at a time is the reliable fix.
Why are some response bodies empty?
Browsers only keep bodies when asked to, and many are dropped for images and large downloads. The viewer shows what the file contains and says when a body is missing, binary or compressed.
Can I use it with HAR files from Firefox or Safari?
HAR 1.2 files from other tools generally open, because the viewer reads the standard fields and tolerates missing ones. Fields only Chrome writes, such as the initiator, are shown when present.
The free plan covers everything on this page. HAR Desk Pro ($29, paid once) is described on the HAR Desk page.
