Multi-source search that fans out to ALL available providers (Tavily, Perplexity, Gemini, Exa) in parallel and returns provider-attributed results. USE WHEN user asks to search, research, look up, find information, check current events, compare topics, get citations, semantic search, or needs web-grounded answers. ALWAYS use this skill for any web search — even simple lookups benefit from cross-provider verification. The default mode searches ALL providers simultaneously. Do not pick just one source when you have four.
From search-hubnpx claudepluginhub lifegenieai/lifegenie-claude-marketplace --plugin search-hubThis skill is limited to using the following tools:
Guides Next.js Cache Components and Partial Prerendering (PPR) with cacheComponents enabled. Implements 'use cache', cacheLife(), cacheTag(), revalidateTag(), static/dynamic optimization, and cache debugging.
Migrates code, prompts, and API calls from Claude Sonnet 4.0/4.5 or Opus 4.1 to Opus 4.5, updating model strings on Anthropic, AWS, GCP, Azure platforms.
Proposes cuts, reorganization, and simplification to improve document structure, clarity, and flow while preserving comprehension. Use for structural or editorial reviews.
Multi-source search that fans out to all available providers in parallel. The goal is information density and signal, not cost optimization. A single source when you have four available is a failure to check.
Every search hits ALL available providers simultaneously. Each provider sees the web differently — Perplexity synthesizes with citations, Tavily returns raw results with snippets, Gemini uses Google's live index, and Exa does neural semantic matching. Cross-referencing what multiple providers surface for the same query reveals what's actually important versus what's noise.
# Default: multi-source across all available providers
bun ${CLAUDE_PLUGIN_ROOT}/tools/search.ts "<query>"
# Subset of providers
bun ${CLAUDE_PLUGIN_ROOT}/tools/search.ts "<query>" --providers tavily,perplexity
# Single provider (only when explicitly needed)
bun ${CLAUDE_PLUGIN_ROOT}/tools/search.ts perplexity "<query>" --model sonar-pro
After running a multi-source search, present results organized BY PROVIDER so the user can see what each source independently found. This is the required output format:
### Google (Gemini)
[Gemini's answer reproduced faithfully — see rules below]
### Tavily
[Tavily's answer reproduced faithfully — see rules below]
### Perplexity
[Perplexity's answer reproduced faithfully — see rules below]
### Exa
[Exa's results reproduced faithfully — see rules below]
### Cross-Provider Signal
[YOUR synthesis goes here — this is the ONLY section where you add interpretation]
Each provider returns an answer field and/or results array. Your job in the
per-provider sections is to faithfully reproduce what the provider actually
said — not to rewrite, summarize, editorialize, or "improve" it.
answer field: reproduce it verbatim or
near-verbatim. You may lightly format for readability (add line breaks, bullet
points) but do NOT change the substance, reorder claims, drop facts, or inject
your own interpretation. The user needs to see what the provider actually
said, not your version of it.results (no answer): list the top results
with their titles, snippets, and URLs. Do not blend them into a narrative.The reason this matters: the whole point of multi-source search is to let the user compare what different providers independently found. If you rewrite their answers, you destroy the signal — the user can't tell what came from where, and can't judge source reliability. Provider sections are evidence; your job is to present the evidence cleanly, not to be the witness.
This is the ONLY section where you add your own analysis:
| Situation | Mode | Why |
|---|---|---|
| Any research question | multi (default) | Cross-reference for signal |
| Quick factual lookup | multi (default) | Still benefits from verification |
| User says "search for X" | multi (default) | Always multi unless told otherwise |
| User explicitly names a provider | single | "ask Perplexity about..." |
| Deep research with Perplexity | single | sonar-deep-research is its own workflow |
| Semantic similarity search | single exa | Exa's neural search is specialized |
The bias is always toward multi. Only go single-provider when the user explicitly requests it or the task is inherently single-provider (like Exa similarity search).
Each provider brings something different to the table:
| Provider | Strength | Returns |
|---|---|---|
| Gemini | Google's live index, free | Grounded answer + source URLs |
| Tavily | Fast web results, free tier | Ranked results with snippets |
| Perplexity | Cited synthesis, deep research | Answer with inline citations |
| Exa | Neural/semantic matching | Semantically relevant results |
When targeting a single provider, these options are available:
--model sonar|sonar-pro|sonar-reasoning-pro|sonar-deep-research--depth basic|advanced, --max-results N--type fast|auto|deep, --category news|"research paper"|company,
--answer--max-results Nbun ${CLAUDE_PLUGIN_ROOT}/tools/search.ts --status
Reports which providers are configured and available before running queries.