Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Summarize newsletters and articles, extract key insights, create reading lists
.claude/skills/sundial-org-newsletter-digest/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-15 | ✗→✓ | ▲ Improved | 102% | 0% |
| case-21 | ✗→✓ | ▲ Improved | 59% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 13% | 0% |
| case-05 | ✓→✓ | ＝ Same ✓ | 144% | 0% |
| case-19 | ✓→✓ | ＝ Same ✓ | 19% | 0% |
Tame your inbox. Get AI summaries of newsletters and articles so you read what matters.
Processes forwarded newsletters and linked articles, extracts key insights, creates prioritized reading lists, and learns your interests over time. Never miss important content, never drown in noise.
Summarize content:
"Summarize this newsletter" [forward/paste]
"Digest https://article-url.com"
"What's important in my newsletters today?"Build reading lists:
"Add to reading list: [url]"
"What's on my reading list?"
"Clear completed reads"Get insights:
"Key takeaways from tech newsletters this week"
"Trending topics across my subscriptions"
"Compare viewpoints on [topic]"Manage preferences:
"I'm interested in AI, fintech, productivity"
"Skip newsletters about marketing"
"Prioritize Stratechery"| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 6,169 | 5,457 | -12% | 1 | 1 | 0% | 908 | 1,063 | +17% | 0 | 0 | — |
case-02 | fail→fail | 6,892 | 4,653 | -32% | 1 | 1 | 0% | 657 | 1,064 | +62% | 0 | 0 | — |
case-03 | pass→pass | 4,403 | 3,217 | -27% | 1 | 1 | 0% | 665 | 753 | +13% | 0 | 0 | — |
case-04 | fail→fail | 5,744 | 3,651 | -36% | 1 | 1 | 0% | 923 | 833 | -10% | 0 | 0 | — |
case-05 | pass→pass | 2,514 | 3,049 | +21% | 1 | 1 | 0% | 287 | 701 | +144% | 0 | 0 | — |
case-06 | fail→fail | 5,775 | 4,842 | -16% | 1 | 1 | 0% | 824 | 977 | +19% | 0 | 0 | — |
case-07 | fail→fail | 11,954 | 4,987 | -58% | 1 | 1 | 0% | 1,680 | 950 | -43% | 0 | 0 | — |
case-08 | fail→fail | 2,206 | 3,060 | +39% | 1 | 1 | 0% | 320 | 734 | +129% | 0 | 0 | — |
case-09 | fail→fail | 14,957 | 9,227 | -38% | 1 | 1 | 0% | 2,065 | 1,510 | -27% | 0 | 0 | — |
case-10 | fail→fail | 16,710 | 14,914 | -11% | 1 | 1 | 0% | 2,329 | 2,240 | -4% | 0 | 0 | — |
case-19 | pass→pass | 5,452 | 5,561 | +2% | 1 | 1 | 0% | 897 | 1,064 | +19% | 0 | 0 | — |
case-11 | fail→fail | 4,004 | 8,230 | +106% | 1 | 1 | 0% | 509 | 1,460 | +187% | 0 | 0 | — |
case-12 | fail→fail | 4,505 | 2,430 | -46% | 1 | 1 | 0% | 664 | 656 | -1% | 0 | 0 | — |
case-13 | fail→fail | 6,491 | 5,079 | -22% | 1 | 1 | 0% | 874 | 1,012 | +16% | 0 | 0 | — |
case-14 | fail→fail | 5,884 | 4,292 | -27% | 1 | 1 | 0% | 783 | 902 | +15% | 0 | 0 | — |
case-15 | fail→pass | 7,772 | 13,979 | +80% | 1 | 1 | 0% | 1,157 | 2,338 | +102% | 0 | 0 | — |
case-16 | fail→fail | 2,632 | 3,034 | +15% | 1 | 1 | 0% | 327 | 708 | +117% | 0 | 0 | — |
case-17 | fail→fail | 7,438 | 4,811 | -35% | 1 | 1 | 0% | 1,194 | 951 | -20% | 0 | 0 | — |
case-18 | fail→fail | 5,692 | 5,025 | -12% | 1 | 1 | 0% | 905 | 1,089 | +20% | 0 | 0 | — |
case-20 | fail→fail | 11,975 | 7,480 | -38% | 1 | 1 | 0% | 1,824 | 1,397 | -23% | 0 | 0 | — |
case-21 | fail→pass | 11,312 | 14,203 | +26% | 1 | 1 | 0% | 1,672 | 2,654 | +59% | 0 | 0 | — |
case-22 | fail→fail | 6,111 | 2,642 | -57% | 1 | 1 | 0% | 763 | 723 | -5% | 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. The headline lift of +9 percentage points is the difference between those two pass rates over the 22 comparable cases.
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.