Recorded on September 25, 2026
Chrono City: from brief to verified pull request
A five-era city scene looked ready after the first build. Browser QA found two unmet criteria, sent the work back and verified the fixes before the pull request was ready. This is the record of that run.
The prompt comes from Bijan Bowen, who uses it to test AI models. The screenshots are the QA agent's own captures from its running browser.
- To the pull request
- 1 h 48 min
- Agent runs
- 6
- Criteria verified
- 11 of 11
- Model spend
- $1.00
What was asked
Build an interactive city block that moves between 1945, 1965, 1985, 2005 and 2025. Visitors must be able to select each era and open period-aware details for scene objects.
What happened on the way
In the first delivery cycle, the year label switched to 1985 but the city did not render. QA reproduced the blank purple viewport three times. It also found empty descriptions in the info cards. Both findings went back to the developer agent before review.
The first build looked done. QA clicked through it anyway.
In an agent session, you are the QA person for your own automation. You open the browser, click every year, spot what is broken and write the follow-up prompt, round after round. Here the QA agent did that work and sent the build back before anyone was asked to review it.
Acceptance criterion 3
One of the five eras rendered nothing

“Selecting 1985 renders no city: the viewport is a flat purple film-grain field while the HUD reports 1985. Reproduced three times.”

1985 renders its full neon skyline, whether it is picked on the slider or with the 3 key.
Acceptance criterion 6
The info cards were empty

“Clicking a building flies the camera in and opens the card, but the card contains no description.”

Every building, vehicle and pedestrian card carries period copy, and a new test fails if one is ever empty again.
From task to pull request in 1 h 48 min
- Architect11:14 AM · 11m 22s101.2K tokens
- Developer11:25 AM · 38m 1s21.4M tokens
- QA12:04 PM · 16m 50s4M tokensSent back
- Developer12:21 PM · 14m 52s6.2M tokens
- QA12:36 PM · 21m 53s2.5M tokens
- Reviewer12:58 PM · 3m 56s443K tokens
- Architect’s estimate
- 12M
- Tokens used
- 34.7M
- Model spend
- $1.00
- Served from cache
- 97%
The architect estimated 12M tokens. The run used 34.7M: the developer checked its own renders, QA exercised every era in a real browser, and one rework round fixed what QA found. With 97% of those tokens served from cache, the whole task cost $1.00.
All five eras, as QA captured them

1945

1965

1985

2005

2025
It takes longer than a demo. That is the point: you get a pull request that has already been tested, not one you still have to test.
The outcome
The architect planned the work, the developer built it twice, QA ran twice and a reviewer checked the result. All 11 acceptance criteria passed in the second cycle. The pull request was ready 1 hour 48 minutes after the task was created.
The record
- 1 task recorded in CoRoid