Skip to content

valyuAI/skills

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 

Repository files navigation

Valyu Agent Skills

Agent skills for integrating Valyu APIs into AI agents and assistants.

These skills follow the Agent Skills specification, making them compatible with Claude Code, Claude Desktop, and other agent platforms.

What is Valyu?

Valyu provides a unified API for AI agents to access real-time information across:

  • Web - General internet content
  • Academic - arXiv, PubMed, bioRxiv, medRxiv (40M+ papers)
  • Financial - SEC filings, earnings, stock data, crypto
  • Healthcare - Clinical trials, drug labels, ChEMBL, DrugBank
  • Economic - FRED, BLS, World Bank data
  • News - Real-time news with date/country filtering
  • Predictions - Polymarket, Kalshi odds
  • Transportation - UK Rail, ship tracking
  • Patents - US patent database

Available Skills

Skill Description
valyu-search Complete Valyu API toolkit - search, content extraction, AI answers, deep research

Quick Start

For AI Agents

Installation

npx skills add valyuAI/skills

The skill provides instructions for:

  • Search API - Find information across 25+ data sources
  • Contents API - Extract clean markdown from URLs
  • Answer API - Get AI-synthesized answers with citations
  • DeepResearch API - Generate comprehensive research reports

For Developers

  1. Get a Valyu API key at platform.valyu.ai ($10 free credits)

  2. Install the SDK:

    npm install valyu-js
    # or
    pip install valyu
  3. Use the API:

    import Valyu from 'valyu-js';
    
    const valyu = new Valyu("your-api-key");
    
    // Search across all sources
    const results = await valyu.search({
      query: "transformer architecture attention mechanism 2024",
      searchType: "all",
      maxNumResults: 10
    });
    
    // Get AI-powered answer
    const answer = await valyu.answer({
      query: "What are the latest developments in quantum computing?"
    });
    
    // Extract content from URL
    const content = await valyu.contents({
      urls: ["https://example.com/article"],
      summary: "Extract key findings in 3 bullet points"
    });
    
    // Deep research (async)
    const task = await valyu.deepResearch.create({
      query: "AI chip market competitive landscape 2024",
      model: "standard"  // fast (~5min), standard (~15min), heavy (~90min)
    });

Project Structure

valyu-agent-skills/
├── README.md
└── valyu-search/
    ├── scripts/                    # CLI tools
    └── best-practices/
        ├── SKILL.md                # Main skill file (start here)
        └── references/
            ├── api-guide.md        # Complete API documentation
            ├── recipes.md          # Index of all 27 recipes
            ├── datasources.md      # Available data sources
            ├── prompting.md        # Query writing best practices
            ├── design-philosophy.md
            ├── search-recipes/     # 14 search patterns
            ├── content-recipes/    # 6 content extraction patterns
            ├── answer-recipes/     # 4 answer generation patterns
            ├── deepresearch-recipes/  # 3 research depth levels
            └── integrations/       # 12 platform guides

API Decision Tree

What do you need?

├─ Find information across multiple sources
│  └─ Search API (/v1/search)
│
├─ Extract content from specific URLs
│  └─ Contents API (/v1/contents)
│
├─ Get an AI-synthesized answer with citations
│  └─ Answer API (/v1/answer)
│
├─ Generate a comprehensive research report
│  └─ DeepResearch API (/v1/deepresearch)
│
└─ Discover available data sources
   └─ Datasources API (/v1/datasources)

Recipes

Search Recipes

Recipe Use Case
basic-search-all Search across all sources
academic-search Research papers (arXiv, PubMed)
finance-search SEC filings, earnings, stocks
news-search Real-time news with filtering
healthcare-and-bio-search Clinical trials, drug data

Content Recipes

Recipe Use Case
basic-content-extraction Clean markdown from URLs
structured-extraction Extract data via JSON schema

Answer Recipes

Recipe Use Case
basic-answer AI answers with citations
answer-with-streaming Progressive response generation

DeepResearch Recipes

Recipe Duration Use Case
fast-research ~5 min Quick lookups
standard-research ~15 min Standard research
heavy-research ~90 min Comprehensive analysis

Integration Guides

Platform Guide
Anthropic Claude integrations/anthropic.md
OpenAI integrations/openai.md
Vercel AI SDK integrations/vercel-ai-sdk.md
LangChain integrations/langchain.md
LlamaIndex integrations/llamaindex.md
MCP Server integrations/mcp-server.md

Query Best Practices

Keep queries under 400 characters. Be specific, not verbose.

BAD:  "I want to know about AI"
GOOD: "transformer attention mechanism survey 2024"

BAD:  "Apple financial information"
GOOD: "Apple revenue growth Q4 2024 earnings SEC filing"

Split complex requests into multiple queries:

# Don't do this
"Tesla stock, products, and news"

# Do this instead
Query 1: "Tesla stock performance Q4 2024"
Query 2: "Tesla Cybertruck production updates"
Query 3: "Tesla FSD autonomous driving progress"

See prompting.md for the complete guide.

Links

License

MIT