Skip to main content

Tavily Answer

Overview​

The TavilyAnswerTool is part of the Tavily Search API integration, specifically designed to provide direct answers to queries. This tool is optimized for scenarios where you need concise, accurate responses rather than raw search results.

Installation​

Add these dependencies to your project:

dependencies:
  langchain: { version }
  langchain_community: { version }

Install via terminal:

Dart​

dart pub add langchain langchain_community

Flutter​

flutter pub add langchain langchain_community

Configuration​

Authentication Options​

TavilyAnswerTool(apiKey: 'your_tavily_key')

2. Base URL Override (For proxies/custom endpoints)​

TavilyAnswerTool(baseUrl: 'https://your-proxy.com/')

Basic Usage​

import 'package:langchain_community/langchain_community.dart';

final tool = TavilyAnswerTool(
  apiKey: Platform.environment['TAVILY_API_KEY'],
);

void main() async {
  final answer = await tool.invoke('Explain quantum entanglement simply');
  print(answer);
  // -> 'Quantum entanglement is a phenomenon where two or more particles...'
}

Advanced Usage​

Custom Configuration with Domain Filtering​

final expertTool = TavilyAnswerTool(
  baseUrl: 'https://science-proxy.com/',
  defaultOptions: const TavilyAnswerToolOptions(
    searchDepth: TavilySearchDepth.advanced,
    includeDomains: ['nasa.gov', 'nature.com'],
  ),
);

void main() async {
  final expertAnswer = await expertTool.invoke('Latest Mars rover findings');
  print(expertAnswer);
  // -> The latest findings from NASA's Perseverance Mars rover...
}

Agent Integration Example​

void main() async {
  final aiAgent = ToolsAgent.fromLLMAndTools(
    llm: ChatOpenAI(apiKey: openAiKey),
    tools: [
      TavilyAnswerTool(apiKey: 'tavily_key'),
      CalculatorTool(),
    ],
  );

  final executor = AgentExecutor(agent: aiAgent);

  final res = await executor.run(
    'Calculate the GDP growth rate of France in 2023 using official sources.',
  );

  print(res);
  // -> "The GDP growth rate of France in 2023 was 0.70%, as reported by EUROSTAT.
}

Error Handling​

try {
return await tool.invoke(query);
} on TavilyClientException catch (e) {
print('Error ${e.message}: ${e.code}');
return 'Failed to retrieve data';
}