Acompleted
Saved artifact
3/8checks passed
$0.00000 tokens
unknown
NEW MODEL. OLD WORK. FRESH EVIDENCE.
See what actually
got better.
Feedback search and export — synthetic reference
Source: Deliberately flawed synthetic fixture; not a historical model output.
THE RESULT
Measured improvement
No model was called. This compares a deliberately flawed synthetic app with a hand-authored reference repair.
+5new passes0 regressions
Rcompleted
Hand-authored reference (no model)
8/8checks passed
$0.00000 tokens
none: hand-authored reference
Effective cap $2.0000
Every comparison
| Versions | Measured result | New passes | Regressions |
|---|---|---|---|
| A → R | Measured improvement | 5 | 0 |
Each row reads left → right. Any lost pass takes precedence over new passes. One sample per workflow; there is no automatic winner or adoption.
EXPERIMENT COST
$0.0000 / $2.0000 stop budget
0reported tokens, including cache reads/writes
No model was called. Failed calls count. Provider stops can overshoot by one in-flight response.
Original intent Preserved before the revision
Let a product team search feedback titles or tags without caring about case or surrounding whitespace, combine search with priority filters, and export exactly the visible items as valid CSV that preserves the original fields. Counts and reset must remain truthful.
Explicitly included by the exporter · Intent confirmed: True
01 / BEHAVIOR
The same checks. Every version.
| Frozen acceptance check | A | R |
|---|---|---|
| Check 1 | PASS | PASS |
| Check 2 | PASS | PASS |
| Check 3 | FAIL | PASS |
| Check 4 | FAIL | PASS |
| Check 5 | FAIL | PASS |
| Check 6HOLDOUT | FAIL | PASS |
| Check 7 | PASS | PASS |
| Check 8HOLDOUT | FAIL | PASS |
Check definitions are not sent to the model. Holdout cases require inputs unavailable in shared source/data; ordinary visible inputs are not holdouts. Timings include navigation, actions and assertion waits; these are not performance benchmarks. Hover a result for failure details.
02 / VISIBLE CHANGE
Look at the same moment.
Same viewport for each corresponding check; 1200 × 820 by default. Per-check dimensions are in the evidence JSON. Click any image to inspect it. Visual differences are evidence, not an automatic quality score.
Evidence included in this export.
Run identity, costs and comparison evidence
{
"status": "completed",
"manifest": {
"input_result_sha256": "fb6332679c6c3484bbdd3396369b1081b370bcea68a44f5895735a70f66fb8c2",
"source": "d161974a58956fd915b98dafb0c4924aded8be892d0b34d433e083ebb878ffd0",
"checks": "66b63820c06e197d34f2876a108d07c36ddbd2c418c550cc089344e291e5f08f"
},
"fingerprint": null,
"source_unchanged": true,
"source_context": null,
"model_calls": 0,
"cost": {
"total_usd": 0.0,
"limit_usd": 2.0,
"known_spend_usd": 0.0,
"basis": "No model was called"
},
"comparisons": {
"A_R": {
"verdict": "improved",
"improvements": [
"check-3",
"check-4",
"check-5",
"check-6",
"check-8"
],
"regressions": []
}
}
}Exported summary JSON · Source, raw prompts/receipts, check definitions, private labels and paths are excluded. Metrics are reported from the input result; this export is not independent verification.