Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Web search and content fetching using Linkup extension. Use when needing to search the web, get answers to questions with sources, or fetch content from specific URLs. Provides three tools: linkup_web_search (discovery), linkup_web_answer (direct answers), linkup_web_fetch (URL content extraction).
.claude/skills/dicklesworthstone-linkup/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-13 | ✗→✓ | ▲ Improved | -44% | 0% |
| case-21 | ✓→✗ | ▼ Worse | 100% | 0% |
| case-22 | ✓→✗ | ▼ Worse | 58% | 0% |
| case-20 | ✓→✓ | ＝ Same ✓ | 235% | 0% |
| case-01 | ✗→✗ | ＝ Same ✗ | -60% | 0% |
Web search and content fetching tools powered by Linkup API.
Search the web and get sources with content snippets.
linkup_web_search(query: string, deep?: boolean)query: Be specific and detailed. Include context like dates, locations, company names.deep: Use for complex research requiring multiple searches. Default: false (faster).Use when: Discovering information across multiple sources, researching topics, comparing perspectives.
Get a synthesized answer with source citations.
linkup_web_answer(query: string, deep?: boolean)Use when: Need a direct answer to a specific question, quick facts with citations.
Fetch content from a URL as clean markdown.
linkup_web_fetch(url: string, renderJs?: boolean)url: The URL to fetch.renderJs: Set false for static pages (faster). Default: true.Use when: Reading documentation, following up on search results, extracting content from known URLs.
| Need | Tool | |------|------| | Find information across sources | linkup_web_search | | Get a direct answer with sources | linkup_web_answer | | Read content from a known URL | linkup_web_fetch |
Good queries are specific:
| Bad | Good | |-----|------| | "Microsoft revenue" | "Microsoft fiscal year 2024 total revenue" | | "React hooks" | "React useEffect cleanup function best practices" | | "AI news" | "OpenAI announcements January 2026" |
Add context:
Standard (default): Simple questions, quick lookups, known topics.
Deep: Complex research, multi-step queries, comprehensive coverage needed.
// Standard - one search is enough
linkup_web_search("Node.js 22 release date")
// Deep - needs multiple searches
linkup_web_search("comparison of Rust web frameworks performance benchmarks 2025", deep: true)linkup_web_search to discover sourceslinkup_web_fetch on promising URLs for full contentlinkup_web_answer for direct answer with citationslinkup_web_fetch on known documentation URL/linkup:balance - Check remaining API credits| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 14,979 | 32,718 | +118% | 1 | 1 | 0% | 2,571 | 1,039 | -60% | 0 | 0 | — |
case-02 | fail→fail | 12,159 | 7,112 | -42% | 1 | 1 | 0% | 720 | 1,089 | +51% | 0 | 0 | — |
case-03 | fail→fail | 12,312 | 5,632 | -54% | 1 | 1 | 0% | 2,137 | 910 | -57% | 0 | 0 | — |
case-04 | fail→fail | 3,566 | 4,793 | +34% | 1 | 1 | 0% | 583 | 900 | +54% | 0 | 0 | — |
case-05 | fail→fail | 10,218 | 5,278 | -48% | 1 | 1 | 0% | 1,925 | 934 | -51% | 0 | 0 | — |
case-06 | fail→fail | 17,597 | 5,137 | -71% | 1 | 1 | 0% | 2,907 | 875 | -70% | 0 | 0 | — |
case-07 | fail→fail | 2,580 | 4,110 | +59% | 1 | 1 | 0% | 418 | 816 | +95% | 0 | 0 | — |
case-08 | fail→fail | 37,749 | 6,113 | -84% | 1 | 1 | 0% | 6,735 | 970 | -86% | 0 | 0 | — |
case-09 | fail→fail | 2,854 | 4,775 | +67% | 1 | 1 | 0% | 392 | 877 | +124% | 0 | 0 | — |
case-10 | fail→fail | 17,073 | 6,128 | -64% | 1 | 1 | 0% | 2,622 | 1,002 | -62% | 0 | 0 | — |
case-11 | fail→fail | 18,462 | 5,440 | -71% | 1 | 1 | 0% | 2,704 | 941 | -65% | 0 | 0 | — |
case-12 | fail→fail | 5,427 | 5,028 | -7% | 1 | 1 | 0% | 958 | 852 | -11% | 0 | 0 | — |
case-13 | fail→pass | 9,125 | 1,334 | -85% | 1 | 1 | 0% | 1,459 | 822 | -44% | 0 | 0 | — |
case-14 | fail→fail | 16,107 | 6,919 | -57% | 1 | 1 | 0% | 2,319 | 1,056 | -54% | 0 | 0 | — |
case-15 | fail→fail | 33,025 | 7,717 | -77% | 1 | 1 | 0% | 4,563 | 1,065 | -77% | 0 | 0 | — |
case-16 | fail→fail | 13,880 | 5,079 | -63% | 1 | 1 | 0% | 2,624 | 891 | -66% | 0 | 0 | — |
case-17 | fail→fail | 3,023 | 4,845 | +60% | 1 | 1 | 0% | 469 | 831 | +77% | 0 | 0 | — |
case-18 | fail→fail | 18,316 | 7,084 | -61% | 1 | 1 | 0% | 2,503 | 1,040 | -58% | 0 | 0 | — |
case-19 | fail→fail | 16,113 | 6,710 | -58% | 1 | 1 | 0% | 2,834 | 981 | -65% | 0 | 0 | — |
case-20 | pass→pass | 1,950 | 2,598 | +33% | 1 | 1 | 0% | 291 | 976 | +235% | 0 | 0 | — |
case-21 | pass→fail | 9,252 | 8,311 | -10% | 1 | 1 | 0% | 615 | 1,228 | +100% | 0 | 0 | — |
case-22 | pass→fail | 4,867 | 11,190 | +130% | 1 | 1 | 0% | 871 | 1,377 | +58% | 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 2 counted toward the lift figure. The other 20 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 -5 percentage points is the difference between those two pass rates over the 2 comparable cases. 2 cases got worse with the skill loaded, and they are 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.