Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Summarize a customer interview transcript into a structured template with JTBD, satisfaction signals, and action items. Use when processing interview recordings or transcripts, synthesizing discovery interviews, or creating interview summaries.
.claude/skills/phuryn-summarize-interview/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 20% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 5% | 0% |
| case-15 | ✗→✓ | ▲ Improved | -35% | 0% |
Transform an interview transcript into a structured summary focused on Jobs to Be Done, satisfaction, and action items.
You are summarizing a customer interview for the product discovery of $ARGUMENTS.
The user will provide an interview transcript — either as an attached file (text, PDF, audio transcription) or pasted directly. Read any attached files first.
**Date**: [Date and time of the interview]
**Participants**: [Full names and roles]
**Background**: [Background information about the customer]
**Current Solution**: [What solution they currently use]
**What They Like About Current Solution**:
- [Job to be done, desired outcome, importance, and satisfaction level]
**Problems With Current Solution**:
- [Job to be done, desired outcome, importance, and satisfaction level]
**Key Insights**:
- [Unexpected findings or notable quotes]
**Action Items**:
- [Date, Owner, Action — e.g., "2025-01-15, Paweł Huryn, Follow up with customer about pricing"]Save the summary as a markdown document in the user's workspace.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 6,127 | 6,137 | +0% | 1 | 1 | 0% | 769 | 1,243 | +62% | 0 | 0 | — |
case-02 | fail→fail | 4,041 | 5,524 | +37% | 1 | 1 | 0% | 329 | 1,499 | +356% | 0 | 0 | — |
case-03 | fail→fail | 11,306 | 5,151 | -54% | 1 | 1 | 0% | 1,259 | 1,197 | -5% | 0 | 0 | — |
case-04 | fail→pass | 5,562 | 4,768 | -14% | 1 | 1 | 0% | 923 | 1,043 | +13% | 0 | 0 | — |
case-05 | fail→pass | 6,084 | 5,646 | -7% | 1 | 1 | 0% | 889 | 1,067 | +20% | 0 | 0 | — |
case-06 | fail→pass | 4,081 | 1,903 | -53% | 1 | 1 | 0% | 722 | 733 | +2% | 0 | 0 | — |
case-07 | pass→pass | 4,231 | 4,396 | +4% | 1 | 1 | 0% | 729 | 1,175 | +61% | 0 | 0 | — |
case-08 | pass→pass | 11,588 | 11,901 | +3% | 1 | 1 | 0% | 1,582 | 1,930 | +22% | 0 | 0 | — |
case-09 | pass→pass | 8,063 | 8,617 | +7% | 1 | 1 | 0% | 1,312 | 1,957 | +49% | 0 | 0 | — |
case-10 | fail→pass | 14,926 | 8,228 | -45% | 1 | 1 | 0% | 1,785 | 1,879 | +5% | 0 | 0 | — |
case-11 | pass→pass | 4,482 | 4,336 | -3% | 1 | 1 | 0% | 693 | 1,191 | +72% | 0 | 0 | — |
case-12 | fail→fail | 3,910 | 3,731 | -5% | 1 | 1 | 0% | 614 | 990 | +61% | 0 | 0 | — |
case-13 | pass→fail | 12,550 | 4,780 | -62% | 1 | 1 | 0% | 1,920 | 1,032 | -46% | 0 | 0 | — |
case-14 | pass→pass | 4,596 | 2,501 | -46% | 1 | 1 | 0% | 775 | 851 | +10% | 0 | 0 | — |
case-15 | fail→pass | 10,799 | 3,544 | -67% | 1 | 1 | 0% | 1,516 | 978 | -35% | 0 | 0 | — |
case-16 | pass→pass | 6,277 | 3,851 | -39% | 1 | 1 | 0% | 660 | 1,041 | +58% | 0 | 0 | — |
case-17 | pass→pass | 2,052 | 3,817 | +86% | 1 | 1 | 0% | 352 | 1,159 | +229% | 0 | 0 | — |
case-18 | pass→pass | 7,665 | 4,200 | -45% | 1 | 1 | 0% | 876 | 1,188 | +36% | 0 | 0 | — |
case-19 | pass→pass | 4,587 | 3,885 | -15% | 1 | 1 | 0% | 779 | 778 | -0% | 0 | 0 | — |
case-20 | pass→pass | 12,768 | 8,452 | -34% | 1 | 1 | 0% | 2,260 | 1,969 | -13% | 0 | 0 | — |
case-21 | fail→fail | 3,508 | 4,034 | +15% | 1 | 1 | 0% | 779 | 1,067 | +37% | 0 | 0 | — |
case-22 | pass→pass | 23,400 | 14,856 | -37% | 1 | 1 | 0% | 3,565 | 3,066 | -14% | 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 21 counted toward the lift figure. The other 1 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 +18 percentage points is the difference between those two pass rates over the 21 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.