Build a law firm's deadline calendar from trigger events with calendar and court-day counting, mail service and court holidays
Exact prompt
# Build a law firm's deadline calendar from trigger events, counting rules, service by mail and court holidays Inputs: `events.csv` (event_id, matter_id, trigger_date, rule_id, service), `rules.json` (each rule's days, direction and counting method) and `court-holidays.json`. Everything is synthetic. `rules.json` and these rules are the record; where `docketing-note.txt` disagrees, the rules win. Rules: - A court day is a date that is not a Saturday, Sunday or court holiday. - calendar counting: start from the trigger date, don't count the trigger day, and count every day; the deadline is the last day counted. If that day is not a court day, move forward to the next court day for an after rule, and back to the previous court day for a before rule. - court_days counting: count only court days (never the trigger day), forward for after rules and backward for before rules; the deadline is the last court day counted. - service mail adds 3 calendar days to an after rule's deadline once it has been computed; if that lands on a day that is not a court day, move forward to the next court day. Service has no effect on before rules. - Each matter's next deadline is its earliest deadline on or after 2026-11-02 (ties: lowest event number); null if it has none. - Write `out/calendar.json`: `{"deadlines": {"<event_id>": "<YYYY-MM-DD>", ...}, "next_by_matter": {"<matter_id>": {"event_id": <id>, "date": "<YYYY-MM-DD>"} or null, ...}}`. ## How this run works - Work on your own: nobody will answer questions in this task, and writing to `ask/` counts against the run. - Write only under `out/`. Create `out/` if it is missing. Inputs are synthetic; don't fetch anything from the network. - Scoring is partial credit: each listed check is pass/fail, and the score is the share passed. Numbers are compared to within 0.005 unless the task says otherwise.
Compute every deadline in a synthetic firm's docket from its trigger events. Calendar-day rules exclude the trigger day and roll forward for after-deadlines but backward for before-deadlines, court-day rules skip weekends and court holidays (including the Thanksgiving pair and Christmas Eve), mail service adds three days to after-deadlines only and rolls again, and each matter's next deadline on or after the review date is picked with ties by event number. A docketing note gets the day-one, roll direction and mail rules wrong. Original benchmark task written for usecaserank.com (not based on a specific public post). Headless, offline run on a reproducible fixture. Each run's notes say how its cost was measured (estimates are marked).
Success criteria: calendar_after before_rules court_days mail_service next_by_matter
Best model for this task so far: Claude Haiku 5.5 (effort default (no --effort flag) · thinking on) (success, $0.0052, 21s).
All runs (3)
100% pass rate across 3 eligible runs: 3 pass, 0 partial, 0 fail. Partial and failed runs are listed below with everything else.ⓘ
| Model | Cost | Time | Step-insⓘ | Autonomy | Outcome | Trustⓘ | Date | Compare |
|---|---|---|---|---|---|---|---|---|
| AClaude Opus 5.5default · thinking on | $0.11 | 21s | 0 | Fully autonomous | pass | Trust 80% | 2026-10-24 | |
| AClaude Haiku 5.5default · thinking on | $0.0052 | 21s | 0 | Fully autonomous | pass | Trust 80% | 2026-10-24 | |
| AClaude Sonnet 5.5default · thinking on | $0.06 | 13s | 0 | Fully autonomous | pass | Trust 80% | 2026-10-24 |
Embed this result
A live badge for this use case. It links back here and updates as new runs are published.
Get new runs and ranking changes weekly
Which models got better or cheaper on real tasks like this one.