perplexity-research
par hushenglang
Web & Frontend DevelopmentRésumé
Conduct deep research using Perplexity Agent API with web search, reasoning, and multi-model analysis.
Description
Ce que fait ce skill
Conducts web-backed research using the Perplexity Agent API, combining live search with configurable reasoning depth. Supports single queries, multi-model comparison, streaming responses, and multi-turn conversations. Tracks token usage and cost per request.
Combining live web search with adjustable reasoning effort in one client lets you tune between speed and depth without switching tools.
Quand l'utiliser
- Researching current market conditions or competitor activity
- Validating findings by comparing answers across multiple LLMs
- Analyzing investment risks with real-time web data
- Tracking emerging trends in a fast-moving industry
- Getting a quick answer to a time-sensitive factual question
Exemple de workflow
User asks: Analyze recent developments in the AI chip market
- 1
Initialize PerplexityClient with PERPLEXITY_API_KEY
- 2
Call research_query() with reasoning_effort set to high
- 3
Retrieve answer, token count, and cost from result
- 4
Optionally call compare_models() across GPT, Claude, and Gemini to validate key findings
- 5
Return synthesized answer with cost summary
A detailed market analysis with sourced web data, cross-model validation, and per-request cost breakdown
Prérequis
- PERPLEXITY_API_KEY environment variable
Installation
clawhub install hushenglang/perplexity-researchExécute la commande CLI, ou colle l'instruction dans Claude Code, Cursor, Codex ou l'agent de code que tu utilises.