Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Free AI search via Exa MCP. Web search for news/info, code search for docs/examples from GitHub/StackOverflow, company research for business intel. No API key needed.
.claude/skills/exa-web-search-free/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -14% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -45% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -3% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -14% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -52% | 0% |
Neural search for web, code, and company research. No API key required.
Verify mcporter is configured:
bashmcporter list exa
If not listed:
bashmcporter config add exa https://mcp.exa.ai/mcp
Search web for current info, news, or facts.
bashmcporter call 'exa.web_search_exa(query: "latest AI news 2026", numResults: 5)'
Parameters:
query - Search querynumResults (optional, default: 8)type (optional) - "auto", "fast", or "deep"Find code examples and docs from GitHub, Stack Overflow.
bashmcporter call 'exa.get_code_context_exa(query: "React hooks examples", tokensNum: 3000)'
Parameters:
query - Code/API search querytokensNum (optional, default: 5000) - Range: 1000-50000Research companies for business info and news.
bashmcporter call 'exa.company_research_exa(companyName: "Anthropic", numResults: 3)'
Parameters:
companyName - Company namenumResults (optional, default: 5)Six additional tools available by updating config URL:
web_search_advanced_exa - Domain/date filtersdeep_search_exa - Query expansioncrawling_exa - Full page extractionpeople_search_exa - Professional profilesdeep_researcher_start/check - AI research agentEnable all tools:
bashmcporter config add exa-full "https://mcp.exa.ai/mcp?tools=web_search_exa,web_search_advanced_exa,get_code_context_exa,deep_search_exa,crawling_exa,company_research_exa,people_search_exa,deep_researcher_start,deep_researcher_check" # Then use: mcporter call 'exa-full.deep_search_exa(query: "AI safety research")'
type: "fast" for quick lookup, "deep" for thorough researchtokensNum (1000-2000) for focused, higher (5000+) for comprehensive| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 4,493 | 7,770 | +73% | 1 | 1 | 0% | 203 | 971 | +378% | 0 | 0 | — |
case-02 | fail→fail | 5,646 | 4,572 | -19% | 1 | 1 | 0% | 306 | 914 | +199% | 0 | 0 | — |
case-03 | fail→fail | 6,564 | 4,632 | -29% | 1 | 1 | 0% | 417 | 855 | +105% | 0 | 0 | — |
case-04 | fail→pass | 8,785 | 4,382 | -50% | 1 | 1 | 0% | 1,495 | 1,284 | -14% | 0 | 0 | — |
case-05 | fail→pass | 11,318 | 3,088 | -73% | 1 | 1 | 0% | 2,299 | 1,265 | -45% | 0 | 0 | — |
case-06 | fail→pass | 5,445 | 1,523 | -72% | 1 | 1 | 0% | 983 | 956 | -3% | 0 | 0 | — |
case-07 | fail→pass | 6,587 | 2,611 | -60% | 1 | 1 | 0% | 1,335 | 1,147 | -14% | 0 | 0 | — |
case-08 | fail→pass | 11,427 | 1,986 | -83% | 1 | 1 | 0% | 2,226 | 1,077 | -52% | 0 | 0 | — |
case-09 | fail→pass | 4,931 | 2,053 | -58% | 1 | 1 | 0% | 947 | 1,038 | +10% | 0 | 0 | — |
case-10 | fail→fail | 3,484 | 1,806 | -48% | 1 | 1 | 0% | 578 | 950 | +64% | 0 | 0 | — |
case-11 | fail→pass | 11,698 | 1,827 | -84% | 1 | 1 | 0% | 1,613 | 1,001 | -38% | 0 | 0 | — |
case-12 | fail→pass | 12,339 | 3,124 | -75% | 1 | 1 | 0% | 2,259 | 1,361 | -40% | 0 | 0 | — |
case-13 | fail→pass | 11,400 | 2,557 | -78% | 1 | 1 | 0% | 2,059 | 1,149 | -44% | 0 | 0 | — |
case-14 | fail→pass | 3,234 | 1,572 | -51% | 1 | 1 | 0% | 551 | 929 | +69% | 0 | 0 | — |
case-15 | fail→pass | 6,778 | 1,679 | -75% | 1 | 1 | 0% | 1,221 | 936 | -23% | 0 | 0 | — |
case-16 | fail→pass | 3,112 | 1,573 | -49% | 1 | 1 | 0% | 497 | 923 | +86% | 0 | 0 | — |
case-17 | fail→pass | 18,953 | 2,140 | -89% | 1 | 1 | 0% | 3,544 | 1,010 | -72% | 0 | 0 | — |
case-18 | fail→pass | 10,239 | 1,997 | -80% | 1 | 1 | 0% | 1,861 | 1,046 | -44% | 0 | 0 | — |
case-19 | fail→pass | 4,286 | 1,416 | -67% | 1 | 1 | 0% | 670 | 904 | +35% | 0 | 0 | — |
case-20 | fail→fail | 8,756 | 6,577 | -25% | 1 | 1 | 0% | 1,834 | 2,064 | +13% | 0 | 0 | — |
case-21 | fail→fail | 4,044 | 4,719 | +17% | 1 | 1 | 0% | 914 | 1,638 | +79% | 0 | 0 | — |
case-22 | fail→fail | 7,052 | 4,579 | -35% | 1 | 1 | 0% | 1,535 | 1,677 | +9% | 0 | 0 | — |
case-23 | fail→pass | 6,402 | 1,724 | -73% | 1 | 1 | 0% | 1,104 | 987 | -11% | 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. 23 cases were attempted, and 20 counted toward the lift figure. The other 3 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 +70 percentage points is the difference between those two pass rates over the 20 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.