Test results
Event to ICS Calendar File: test results
Tested 2026-10-08, skill version 1.0.0 at the time of loading this page. We run every skill on a strong and a weak model before it is listed, and publish both verdicts, including where the weak one fails.
Verdicts
- Date
- 2026-10-08
- Strong · claude-sonnet-5-5 (Claude Code alias "sonnet")
- Wrote a valid iCalendar file for all 24 invitations, checked by parsing each answer: a VTIMEZONE block for every named time zone (Sofia, New York), all-day events ending on the day after the last day, no end time when the text gives none, commas, semicolons and a backslash escaped in the title, a weekly rule with a UTC end date, a relative date worked out from the date given, an offset converted to UTC, a Bulgarian and a German invitation, and no attendee added when the text asked for ten.
- Weak · claude-haiku-5-5 (Claude Code alias "haiku")
- Also 24 of 24 with the skill, including the four cases that need a time zone definition block and the weekly rule with a UTC end date. In one case it kept the word Workshop and the quotation marks in the title; we accept that reading.
With and without the skill
Tested 2026-10-08.
| Sonnet | Haiku | |||
|---|---|---|---|---|
| with | without | with | without | |
| Valid files with the right values (24 invitations) | 24/24 | 24/24 | 24/24 | 19/24 |
The same request on both sides; a fence around the answer is removed first. Sonnet needs no help on these cases. Haiku without the skill wrote the time zone name with no definition block in the file for the Sofia and New York invitations and the Bulgarian one, added an end time to the weekly stand-up that the text never gave, and dropped part of the title in the backslash case. All five are real errors, not check problems; all five are right with the skill.
Same cases and the same checks with and without the skill. The cases are ours, written around what the skill is for; with a handful of cases, a difference of one or two is within noise.
Note
Twenty-four invitations written by us (seven simple, seventeen with a trap), one in Bulgarian and one in German. Each answer was parsed as iCalendar and checked for structure (required properties, DTSTAMP in UTC, a VTIMEZONE for every time zone used, DTEND after DTSTART, never DTEND with DURATION) and for the values the text implies (start time and zone, end, title, location, repeat rule, and properties that must be absent). One run per model and invitation. We relaxed one check after reading Haiku's answer: the title of the backslash case is accepted with or without the word Workshop around it. Without the skill, Sonnet already writes valid files in all of these cases; the gain is for the smaller model.
What was not measured
- Models other than the two named above were not run.
- Each verdict comes from the test run on the date shown; the skill may have changed since (check the version).
- Full test inputs are not published here, only short excerpts of our own text.
- Results on your own texts, languages and domains can differ.