Table of Contents

Class WebReadTool

Namespace
LMKit.Agents.Tools.BuiltIn.Net
Assembly
LM-Kit.NET.dll

Reads the main content of a public web page as clean Markdown: the step after web_search, which only names pages. Navigation, scripts, forms, banners, and repeated site chrome are stripped; the result carries the page title, the canonical URL, the source domain, and the publication date when the page states one.

The model-facing schema is the URL alone. Every limit (timeout, redirects, response size, content cap) and the whole egress policy (public-web or allowed-hosts mode, private-address protection, pinned connections) are server-side settings on WebReadTool.Options, never the model's to choose. 

public sealed class WebReadTool : IBuiltInTool, ITool, IToolMetadata
Inheritance
WebReadTool
Implements
Inherited Members

Examples

Search, then read the winning page:

using LMKit.Agents;
using LMKit.Agents.Tools.BuiltIn.Net;

var agent = Agent.CreateBuilder(model)
.WithTools(t =>
{
t.Register(new WebSearchTool());
t.Register(new WebReadTool());
})
.Build();


var result = await agent.RunAsync(
"Find the latest release notes for the .NET runtime and summarize the breaking changes.");

Constructors

WebReadTool()

Creates the tool with default options: public-web egress, chat-sized content cap.

WebReadTool(Options)

Creates the tool with the given server-side options.

Properties

Description

Gets a concise description of what the tool does.

InputSchema

Gets the JSON Schema defining the expected input arguments.

Name

Gets the stable, unique identifier for this tool.

Methods

InvokeAsync(string, CancellationToken)

Executes the tool with the specified JSON arguments.

Share