A screen recorder with console logs, including the errors before you hit record
Updated
Midge is a screen recorder with console logs. While you record your screen or take a screenshot in Chrome, it captures every console message the page produces, from console.log to uncaught exceptions and unhandled promise rejections, with stack traces, and lines them up with the video. Because Midge keeps a rolling buffer of recent console output, an error that fired before you pressed record is still in the report.
That second part is what most bug reports miss. By the time you notice something broke, the error has already scrolled by in a DevTools window nobody had open. Midge removes the "can you reproduce it with the console open?" step.
What gets captured
All console levels
Midge records log, info, warn, error and debug calls with their timestamps and the values passed to them. Warnings matter more than people think: a React key warning or a deprecation notice logged right before a crash is often the real clue.
Uncaught errors
Exceptions that nobody catches never go through console.error. They bubble up to the window and the browser prints them in red. Midge listens for those too, so a crash in an event handler shows up even if your app has no logging at all.
Unhandled promise rejections
A rejected promise without a .catch() is one of the most common silent failures in modern front ends. A button "does nothing" because a fetch rejected and nobody handled it. Midge records these rejections and the reason they carry.
Stack traces
For errors, Midge saves the stack trace the browser reports. A typical entry looks like this:
Uncaught (in promise) TypeError: Failed to fetch
at submitOrder (orders.js:88:15)
at HTMLButtonElement.onClick (checkout.js:41:9)
If your production bundle is minified, the stack shows minified file names and positions, exactly as the browser reported them, so keep your source maps handy when you read a report from production.
Captured before you press record
Midge starts listening as each page begins to load, before your app's own scripts run, so even errors thrown during startup are caught. It keeps the last few minutes of console output in memory for every open tab.
When you start a screen recording or take a screenshot, that recent history is attached. In practice, this means you can work normally, notice that something looks wrong, and only then decide to report it. The red error from 20 seconds ago is still on the timeline.
A few things about the buffer:
- It lives only in memory and is cleared when the tab closes.
- Nothing leaves your computer unless you create a report.
- It runs on regular http and https pages. You can add sites where Midge should never run, such as your bank or a password manager.
Console and video on one timeline
Every recording includes console logs, network requests, user actions and device info. In the report, the console sits next to the video, sorted by time. Watch the moment the modal fails to open and the matching TypeError is right there, a few lines below the click that triggered it.
Errors pair naturally with network activity. A 500 from /api/orders followed by Cannot read properties of undefined tells the whole story: the server failed and the client did not handle the failure. Read more on the network logs page.
Secrets in console output
Developers log things they should not, especially during debugging: a token, a full user object, a request URL with an API key in the query string. Midge runs the same redaction rules over console text that it uses for network data. Passwords, card numbers, tokens and cookies are redacted in your browser before anything leaves your device.
For example, this line:
auth ok: eyJhbGciOiJIUzI1NiJ9.eyJzdWIiOiI0MiJ9.abc123 for [email protected]
appears in the report as:
auth ok: [REDACTED] for j***@acme.com
The full rule list, and what redaction does not catch, is on the redaction page.
Midge versus exporting the console by hand
You can save console output from Chrome DevTools yourself; our guide on how to export console logs in Chrome walks through it. It works, but it has gaps. DevTools must be open before the error happens, the export is plain text without the video, and nothing removes secrets from it. If you do export logs or a HAR file manually, run the HAR through the free HAR sanitizer before you attach it to a ticket.
With Midge, the console is captured whether DevTools was open or not, it is synced with what you saw, and it is redacted before upload.
Who reads these logs
QA testers stop pasting partial stack traces into tickets. Support agents can attach real error output to an escalation without knowing what a console is. Developers get the exact error, the stack and the surrounding network calls in one place. If that is your role, see Midge for developers.
Midge is free: unlimited recordings, videos up to 15 minutes, and no watermark.
Frequently asked questions
- Which console messages does Midge capture?
- All five levels: log, info, warn, error and debug. It also records uncaught exceptions and unhandled promise rejections, which never go through console.error, along with their stack traces.
- Can Midge capture an error that happened before I started recording?
- Yes. Midge keeps the last few minutes of console output in memory for each tab, so a recording or screenshot made right after an error still contains it. The buffer is cleared when you close the tab and never leaves your device unless you create a report.
- Are secrets in console messages redacted?
- Yes. JWTs, card numbers, known API key formats, private keys and sensitive URL parameters are removed from console text, and emails and international phone numbers are half-masked.
- Do I need to open DevTools?
- No. Midge captures the console without DevTools being open, and the person who opens your report sees the messages next to the video.