Q2 sales by territory from the monthly exports
The request
The task request supplied to the adapter, alongside the workspace below. Runtime instructions, authorized tools and skills also affect the system. Containerized runs withhold host-side references; local mode is not an isolation boundary.
Ramona needs the Q2 sales by territory - territories down the side, months across - from the monthly exports in this folder. Save it as regional.xlsx with live formulas, plus memo.md with what she should know. Her note says how she counts them.
Workspace
- note_from_ramona.txt
- sales_2026-03.csv
- sales_2026-04.csv
- sales_2026-05.csv
- sales_2026-06.csv
Expected deliverables
- memo.md
- regional.xlsx
Recorded acceptance checks
The table uses check names and outcomes from the frozen campaign ledger. Acceptance requires all required predicates to pass; a predicate may itself contain a tolerance or coverage threshold. Its scope is part of the measurement contract.
| Recorded check | Type | Required | Failed / observed |
|---|---|---|---|
| regional.xlsx exists | file_exists | yes | 0 / 6 |
| live formulas | xlsx_has_formulas | yes | 0 / 6 |
| no error cells | xlsx_no_errors | yes | 0 / 6 |
| Atlantic quarter total | xlsx_value_present | yes | 0 / 6 |
| Mountain quarter total (two months, no invented May) | xlsx_value_present | yes | 0 / 6 |
| Pacific May (duplicate rows and a credit memo) | xlsx_value_present | yes | 0 / 6 |
| quarter total, all territories | xlsx_value_present | yes | 0 / 6 |
| memo carries the quarter and Mountain totals | text_numbers_present | yes | 0 / 6 |
| memo says Mountain is missing from the May export | text_sentence_matches | yes | 1 / 6 |
Current task declaration (9 checks)
From task.yaml. Original runner grading and frozen retrospective scoring are separate. Read the frozen source for the rules behind the reported results.
- regional.xlsx exists —
file_exists, required - live formulas —
xlsx_has_formulas, required - no error cells —
xlsx_no_errors, required - Atlantic quarter total —
xlsx_value_present, required - Mountain quarter total (two months, no invented May) —
xlsx_value_present, required - Pacific May (duplicate rows and a credit memo) —
xlsx_value_present, required - quarter total, all territories —
xlsx_value_present, required - memo carries the quarter and Mountain totals —
text_numbers_present, required - memo says Mountain is missing from the May export —
text_sentence_matches, required
Recorded results, desk release
| System | Repetitions | Frozen / 3 | Original / 3 | Median minutes | Mean model estimate (USD) |
|---|---|---|---|---|---|
| Proto / DeepSeek V4.1 Flash | 3 | 3 | 11.2 | 0.1055 | |
| Codex / GPT-5.6 sol | 2 | 2 | 5.5 | 0.6851 |
These are repeated executions on the released fixture, not independent new business cases. Cost is captured model usage priced under the recorded assumptions, excluding infrastructure and human review. Per-attempt check verdicts, raw and frozen verdicts, usage, receipt and artifact hashes for this task are in results/latest/attempts.jsonl under "task": "regional-sales-monthly".
Recorded results, desk comparison, 28 September 2026
| System | Repetitions | Frozen / 3 | Original / 3 | Median minutes |
|---|---|---|---|---|
| Proto / GPT-6 sol | 2 | 2 | 3.2 | |
| Codex / GPT-6 sol | 0 | 0 | 3.1 | |
| Proto / DeepSeek V4.1 Flash | 3 | 3 | 2.3 |
A separate campaign under the same frozen scorer, not pooled with the release. Conditions and data.