Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Deep search via Perplexity API. Three modes: search (quick facts), reason (complex analysis), research (in-depth reports). Returns AI-grounded answers with citations.
.claude/skills/sundial-org-perplexity-2/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-18 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-01 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -6% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -25% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 47% | 0% |
AI-powered web search with three modes for different depth levels.
bash# Quick search (sonar) - facts, summaries, current events {baseDir}/scripts/search.sh "latest AI news" # Reasoning (sonar-reasoning-pro) - complex analysis, multi-step {baseDir}/scripts/search.sh --mode reason "compare React vs Vue for enterprise apps" # Deep Research (sonar-deep-research) - full reports, exhaustive analysis {baseDir}/scripts/search.sh --mode research "market analysis of AI in healthcare 2025"
| Mode | Model | Best For | Cost | |------|-------|----------|------| | search (default) | sonar-pro | Quick facts, summaries, current events | Low | | reason | sonar-reasoning-pro | Complex analysis, comparisons, problem-solving | Medium | | research | sonar-deep-research | In-depth reports, market analysis, literature reviews | High |
| Flag | Description | Default | |------|-------------|---------| | --mode | search, reason, research | search | | --recency | hour, day, week, month | — | | --domains | Comma-separated domain filter | — | | --lang | Language code (pt, en, es, etc.) | — | | --json | Raw JSON output | off |
bash# Search with recency filter {baseDir}/scripts/search.sh --recency week "OpenAI latest announcements" # Search restricted to specific domains {baseDir}/scripts/search.sh --domains "arxiv.org,nature.com" "transformer architecture advances" # Search in Portuguese {baseDir}/scripts/search.sh --lang pt "inteligência artificial no Brasil" # Deep research with JSON output {baseDir}/scripts/search.sh --mode research --json "enterprise AI adoption trends"
Set PERPLEXITY_API_KEY env var:
bashexport PERPLEXITY_API_KEY="pplx-..."
Use search for everyday queries. Reserve research for when you truly need exhaustive analysis.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-18 | fail→pass | 6,854 | 4,070 | -41% | 1 | 1 | 0% | 1,151 | 1,192 | +4% | 0 | 0 | — |
case-01 | fail→pass | 8,528 | 2,479 | -71% | 1 | 1 | 0% | 1,544 | 1,074 | -30% | 0 | 0 | — |
case-02 | fail→pass | 9,323 | 5,811 | -38% | 1 | 1 | 0% | 1,595 | 1,507 | -6% | 0 | 0 | — |
case-03 | fail→pass | 10,198 | 3,727 | -63% | 1 | 1 | 0% | 1,728 | 1,291 | -25% | 0 | 0 | — |
case-04 | fail→pass | 4,638 | 2,621 | -43% | 1 | 1 | 0% | 730 | 1,076 | +47% | 0 | 0 | — |
case-05 | fail→pass | 4,751 | 2,523 | -47% | 1 | 1 | 0% | 712 | 1,022 | +44% | 0 | 0 | — |
case-06 | fail→pass | 6,746 | 2,106 | -69% | 1 | 1 | 0% | 1,080 | 896 | -17% | 0 | 0 | — |
case-07 | fail→pass | 10,160 | 2,571 | -75% | 1 | 1 | 0% | 1,669 | 1,016 | -39% | 0 | 0 | — |
case-08 | fail→pass | 11,576 | 3,887 | -66% | 1 | 1 | 0% | 1,982 | 1,243 | -37% | 0 | 0 | — |
case-09 | fail→pass | 11,458 | 4,359 | -62% | 1 | 1 | 0% | 1,709 | 1,301 | -24% | 0 | 0 | — |
case-10 | fail→pass | 9,789 | 3,367 | -66% | 1 | 1 | 0% | 1,671 | 1,115 | -33% | 0 | 0 | — |
case-11 | fail→pass | 10,525 | 3,112 | -70% | 1 | 1 | 0% | 1,678 | 1,074 | -36% | 0 | 0 | — |
case-12 | fail→pass | 10,236 | 3,183 | -69% | 1 | 1 | 0% | 1,743 | 1,086 | -38% | 0 | 0 | — |
case-13 | fail→pass | 3,912 | 2,774 | -29% | 1 | 1 | 0% | 590 | 1,055 | +79% | 0 | 0 | — |
case-14 | pass→pass | 3,429 | 1,834 | -47% | 1 | 1 | 0% | 565 | 867 | +53% | 0 | 0 | — |
case-15 | fail→pass | 11,620 | 3,891 | -67% | 1 | 1 | 0% | 1,665 | 1,196 | -28% | 0 | 0 | — |
case-16 | fail→pass | 10,240 | 2,333 | -77% | 1 | 1 | 0% | 1,538 | 921 | -40% | 0 | 0 | — |
case-17 | pass→pass | 5,632 | 3,135 | -44% | 1 | 1 | 0% | 748 | 1,088 | +45% | 0 | 0 | — |
case-19 | fail→pass | 12,650 | 4,567 | -64% | 1 | 1 | 0% | 1,989 | 1,370 | -31% | 0 | 0 | — |
case-20 | pass→pass | 3,415 | 2,710 | -21% | 1 | 1 | 0% | 642 | 1,129 | +76% | 0 | 0 | — |
case-21 | pass→pass | 7,927 | 6,967 | -12% | 1 | 1 | 0% | 1,579 | 2,061 | +31% | 0 | 0 | — |
case-22 | pass→pass | 10,311 | 8,737 | -15% | 1 | 1 | 0% | 1,820 | 2,217 | +22% | 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 +77 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.