Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Create Fantastical calendar events and reminders from Codex on macOS. Use when the user asks Codex to add, schedule, parse, preview, or send a meeting, appointment, task, reminder, time block, daily plan item, 行事曆, 日曆, 會議, 待辦, or 提醒 into Fantastical.
.claude/skills/twhsi-fantastical-calendar/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-05 | ✗→✓ | ▲ Improved | -35% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -39% | 0% |
| case-09 | ✓→✓ | ＝ Same ✓ | -15% | 0% |
| case-10 | ✓→✓ | ＝ Same ✓ | 120% | 0% |
| case-12 | ✓→✓ | ＝ Same ✓ | 160% | 0% |
Use this skill to turn a user's natural-language scheduling request into a Fantastical event or reminder through the local macOS Fantastical app.
Prefer the bundled script over ad hoc AppleScript:
bashpython3 scripts/fantastical_parse.py "Project review on 2026-06-19 at 10:00" --notes "Prep deck before the call"
The script defaults to preview mode: it opens Fantastical with the parsed item but does not add it immediately.
--notes instead of overloading the event title.scripts/fantastical_parse.py from this skill folder.Use preview mode unless the user explicitly asks to add/create/schedule it now or has confirmed the parsed result.
Preview:
bashpython3 scripts/fantastical_parse.py "Deep work on 2026-06-19 from 09:00 to 11:00"
Add immediately:
bashpython3 scripts/fantastical_parse.py "Deep work on 2026-06-19 from 09:00 to 11:00" --add-immediately
Dry run for validation:
bashpython3 scripts/fantastical_parse.py "Deep work on 2026-06-19 from 09:00 to 11:00" --dry-run
Use explicit, Fantastical-friendly wording:
Team sync on 2026-06-19 from 10:00 to 10:30Call Alice on 2026-06-20 at 15:00 for 45 minutesRemind me to submit expenses on 2026-06-21 at 09:00Focus block on 2026-06-18 from 14:00 to 16:00For Chinese user requests, keep the user-facing reply in Chinese, but send Fantastical an explicit mixed or English sentence if it reduces parsing risk.
Read references/fantastical-macos.md only when debugging the app integration, updating the script, or adapting it to another Fantastical version.
The current script uses AppleScript parse sentence first and can fall back to the Fantastical URL scheme when needed.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 5,133 | 10,749 | +109% | 1 | 1 | 0% | 872 | 958 | +10% | 0 | 0 | — |
case-02 | fail→fail | 10,345 | 5,872 | -43% | 1 | 1 | 0% | 1,091 | 1,007 | -8% | 0 | 0 | — |
case-03 | fail→fail | 6,097 | 5,850 | -4% | 1 | 1 | 0% | 1,024 | 1,023 | -0% | 0 | 0 | — |
case-04 | fail→fail | 6,998 | 9,567 | +37% | 1 | 1 | 0% | 1,098 | 1,498 | +36% | 0 | 0 | — |
case-05 | fail→pass | 8,764 | 2,005 | -77% | 1 | 1 | 0% | 1,640 | 1,067 | -35% | 0 | 0 | — |
case-06 | fail→fail | 16,127 | 8,222 | -49% | 1 | 1 | 0% | 1,188 | 1,289 | +9% | 0 | 0 | — |
case-22 | fail→fail | 3,925 | 8,601 | +119% | 1 | 1 | 0% | 716 | 1,246 | +74% | 0 | 0 | — |
case-07 | fail→fail | 7,413 | 8,900 | +20% | 1 | 1 | 0% | 1,262 | 1,438 | +14% | 0 | 0 | — |
case-08 | fail→fail | 6,256 | 8,647 | +38% | 1 | 1 | 0% | 1,136 | 1,309 | +15% | 0 | 0 | — |
case-09 | pass→pass | 8,471 | 7,894 | -7% | 1 | 1 | 0% | 1,608 | 1,368 | -15% | 0 | 0 | — |
case-10 | pass→pass | 3,818 | 4,749 | +24% | 1 | 1 | 0% | 707 | 1,558 | +120% | 0 | 0 | — |
case-11 | fail→fail | 1,857 | 6,284 | +238% | 1 | 1 | 0% | 321 | 1,167 | +264% | 0 | 0 | — |
case-12 | pass→pass | 2,947 | 2,928 | -1% | 1 | 1 | 0% | 430 | 1,117 | +160% | 0 | 0 | — |
case-13 | fail→pass | 10,811 | 2,515 | -77% | 1 | 1 | 0% | 1,960 | 1,189 | -39% | 0 | 0 | — |
case-14 | fail→fail | 7,271 | 5,901 | -19% | 1 | 1 | 0% | 510 | 1,136 | +123% | 0 | 0 | — |
case-15 | fail→fail | 3,666 | 7,552 | +106% | 1 | 1 | 0% | 610 | 1,253 | +105% | 0 | 0 | — |
case-16 | pass→pass | 8,097 | 2,427 | -70% | 1 | 1 | 0% | 1,403 | 1,135 | -19% | 0 | 0 | — |
case-17 | fail→fail | 8,641 | 8,674 | +0% | 1 | 1 | 0% | 1,477 | 1,245 | -16% | 0 | 0 | — |
case-18 | fail→fail | 11,708 | 6,570 | -44% | 1 | 1 | 0% | 1,153 | 1,042 | -10% | 0 | 0 | — |
case-19 | pass→pass | 12,576 | 10,198 | -19% | 1 | 1 | 0% | 2,685 | 2,806 | +5% | 0 | 0 | — |
case-20 | pass→pass | 4,454 | 6,353 | +43% | 1 | 1 | 0% | 796 | 1,869 | +135% | 0 | 0 | — |
case-21 | pass→pass | 12,413 | 8,312 | -33% | 1 | 1 | 0% | 2,214 | 2,179 | -2% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 9 counted toward the lift figure. The other 13 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +9 percentage points is the difference between those two pass rates over the 9 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.