EXAMPLE BASELINE REPORT
You fixed both defects and left the verified control patch unchanged.
This shows what the baseline changes: it selects the first Manual Queue task from observed behavior. It is not a level, percentile, or estimate of overall engineering ability.
Reviews
3/3
Submitted
Tests first
2/3
Before editing
Test runs
5
Across all tasks
Help
1
After a failed run
Manual time
16:03
Context only
What happened on each task
| Patch | Outcome | First move | First run | Edits | Help | Time |
|---|---|---|---|---|---|---|
| Permission precedence | Defect fixed | Tests | 3/5 | 1 | 0 | 05:18 |
| Async ownership | Defect fixed | Edit | 2/3 | 2 | 1 | 07:04 |
| Cache identity | Approved unchanged | Tests | 4/4 | 0 | 0 | 03:41 |
OBSERVED PATTERN
Tests came before the first edit on two of three tasks.
The session shows evidence-first behavior twice, one early edit, and one Coach action after a failing run. Those are observations, not a personality score.
First task in Manual Queue
On the next async task, run the original patch once and name the ownership invariant before changing its guard.
Get a queue based on your own process.
Three patches, including a clean control. The report unlocks only after all three reviews are submitted.