Using Tavily Instead of SearxNG

This guide explains how to configure and use Tavily as a replacement for SearxNG in the Perplexica project.

What is Tavily?

Tavily is an AI-native search engine that provides a powerful API for web searches. It's designed to be more focused on AI applications and provides structured data in its responses.

Advantages of Tavily over SearxNG

AI-Optimized: Tavily is specifically designed for AI applications and provides more relevant results for AI-powered search.
Structured Data: Results come in a well-structured format that's easy to process.
AI-Generated Answers: Tavily can generate concise answers to queries based on search results.
Advanced Search Options: Includes features like search depth control, domain filtering, time-based filtering, and result limiting.

Setup Instructions

Get a Tavily API Key:
- Sign up for an account at Tavily
- Navigate to your dashboard and create an API key
Configure the API Key:
- Open your config.toml file
- Add your Tavily API key to the TAVILY field under [API_ENDPOINTS]:
```
[API_ENDPOINTS]
TAVILY = "your-tavily-api-key"
```
- Alternatively, you can set the TAVILY_API_KEY environment variable
Select Tavily as the Search Engine:
- Open your config.toml file
- Set the ENGINE field under [SEARCH] to "tavily":
```
[SEARCH]
ENGINE = "tavily"
```
- Alternatively, you can set the SEARCH_ENGINE environment variable to "tavily"
Verify Configuration:
- The application will now use Tavily for web searches
- You can verify this by checking the console logs, which will show "Using search engine: tavily" when a search is performed

Usage

The Tavily integration is designed to be a drop-in replacement for SearxNG. All existing search functionality will continue to work as before, but with improved results from Tavily.

API Options

The Tavily implementation supports the following options:

topic: 'general' or 'news' (default: 'general')
search_depth: 'basic' or 'advanced' (default: 'advanced')
chunks_per_source: Number of content chunks to retrieve from each source (1-3, default: 3)
max_results: Maximum number of results to return (default: 10)
time_range: Filter results by time range ('day', 'week', 'month', 'year', 'd', 'w', 'm', 'y')
days: Number of days back from the current date to include (for news topic)
include_answer: Include an AI-generated answer ('basic', 'advanced', or true)
include_raw_content: Include the cleaned and parsed HTML content of each result
include_images: Also perform an image search and include the results
include_image_descriptions: Add descriptive text for each image
include_domains: Array of domains to include in search results
exclude_domains: Array of domains to exclude from search results

Response Format

Tavily returns a structured response with the following fields:

query: The search query that was executed
answer: An AI-generated answer to the query (if requested)
images: List of query-related images (if requested)
results: A list of search results, each containing:
- title: The title of the search result
- url: The URL of the search result
- content: A short description of the search result
- score: The relevance score of the search result
- raw_content: The cleaned and parsed HTML content (if requested)
response_time: Time in seconds it took to complete the request

Troubleshooting

If you encounter any issues:

Verify your Tavily API key is correctly set in config.toml or as an environment variable
Check the application logs for any error messages
Ensure your Tavily account has sufficient credits/quota for API calls

Switching Between Search Engines

You can easily switch between SearxNG and Tavily by changing the ENGINE setting in your config.toml file:

[SEARCH]
ENGINE = "searxng"  # Use SearxNG (default)
# or
ENGINE = "tavily"   # Use Tavily

You can also use the SEARCH_ENGINE environment variable to override this setting.

Reverting to SearxNG

If you need to revert to using SearxNG:

Set the ENGINE field under [SEARCH] in your config.toml file to "searxng":
```
[SEARCH]
ENGINE = "searxng"
```
Ensure your SEARXNG entry is properly configured with a valid SearxNG instance URL

4.4 KiB Raw Blame History