Research the web with Lares

Lares can use search and page-reading providers configured in Router to work with public web content. Search providers find relevant pages, while scrape providers read a specific page for analysis or summarization.

This guide connects SearXNG as a search provider, checks the connection with a small source-finding task, and then uses the results in a larger research request.

Learning objectives

In this guide, you will learn how to:

• Choose between search and scrape providers.
• Configure a web research provider in Router.
• Select a search service in Lares and inspect its sources.

Prerequisites

• Lares prepared with a tool-capable chat model.
• One web research provider. This guide uses SearXNG installed from Market.

Choose a web research provider

Router labels providers by the capability they supply.

Capability	Providers	What Lares can do
search	SearXNG, Serper, Tavily	Find pages and return titles, snippets, and links.
scrape	Firecrawl, Tavily, Jina Reader	Read a provided URL for analysis, summarization, or saving.

SearXNG and Firecrawl run as apps on Olares. Install the app from Market, then copy its endpoint from Olares Settings > Applications > [App name] > Entrances. Serper, Tavily, and Jina Reader require credentials from their providers.

Connect SearXNG in Router

1. Open Olares Settings > Applications > SearXNG > Entrances.

2. Select the SearXNG entrance and copy its Endpoint URL.

3. Open Router and go to Tools > Manage providers.

4. Select SearXNG and enter:

   • Provider name: localsearxng
   • SearXNG instance URL: The endpoint copied in step 2

5. Click Add. The provider appears under Available.

6. Click add to enable it.

   

   The provider appears under Configured.

   

Select the search service in Lares

A provider with the search capability must be selected in Lares. A scrape-only provider is called when you ask Lares to read a specific URL and does not appear as the default search service.

1. Open Lares and go to Settings > Web Search.
2. Click Refresh.
3. Select localsearxng from Default search service.

Check the connection

Start with a small request so you can inspect each returned link:

Use web search to find three relevant public sources about preserving a
personal voice when using AI writing tools. For each source, return the
title, URL, and a one-sentence summary.

Open the links and check that the pages match the summaries. If no links appear, confirm that SearXNG works directly, then refresh the provider list in Lares.

Run a larger research task

Once the source check succeeds, ask Lares to search, read, and synthesize material. State the question, source requirements, and desired output. For example:

Research practical methods for preserving an author's voice when using AI
editing tools. Use at least five public sources, including two first-party
sources from writing or publishing organizations. Compare the methods,
identify tradeoffs, and produce a concise report with inline source links.

Review the cited pages before relying on the report. A successful search connection confirms that Lares can retrieve results, but it does not establish that every generated claim is supported by its source.

Learn more

• Get started with Lares: Prepare the model and run a first read-only task.
• Set up SearXNG: Install and use the search service directly.
• Use Firecrawl: Prepare a page-reading service for website content.