web-search

Stats

33 Downloads

Last updated

Project Files

src

test

.gitignore

31 B

LICENSE

1 KB

manifest.json

110 B

package-lock.json

29 KB

package.json

754 B

README.md

3 KB

tsconfig.json

359 B

README

LM Studio Web Search Plugin

A TypeScript-based web search plugin for LM Studio. This plugin enables local LLMs to access the live web, read page content, and perform pattern matching within web pages.

Features

Live Web Search: Query the web using DuckDuckGo (no API key required).
Page Reading: Extract clean, readable text from any URL.
In-Page Search: Find specific text or regex-style patterns within a web page.
Domain Filtering: Restrict searches and page reads to specific allowed domains.
Configurable Context: Adjust the amount of data returned to the model (low, medium, or high).
Today's Date: Get the current local date and time.
Privacy Focused: Operates locally as an LM Studio plugin.

Tools Included

The plugin registers four main tools with LM Studio:

search: Performs web searches. Supports single queries or batches of up to 5 queries.
read: Opens a URL and extracts its text content, automatically stripping HTML tags, scripts, and styles.
find_in_page: Searches for a specific pattern (string or regex) within a given URL and returns the surrounding context.
today: Returns the current local date and time.

Configuration

You can configure the following settings in LM Studio:

Search Context Size: Controls the default search breadth and page extraction size (low, medium, high).
Allowed Domains: An optional allowlist of domains for search results and page fetches.
Location Settings: Set your country, region, city, and timezone to improve search relevance.
Search Results Per Page: Customize the number of results returned (1-10).
Safe Search: Set the safety level for search results (strict, moderate, off).

Installation & Development

Prerequisites

Node.js (v18 or later recommended)
LM Studio installed

Setup

Clone this repository to your local machine.
Install dependencies:
```
npm install
```

Development Commands

Run in Development Mode:
```
npm run dev
```
This uses lms dev to load the plugin into LM Studio for testing.
Build the Plugin:
```
npm run build
```
Compiles the TypeScript source code into the dist directory.
Run Tests:
```
npm test
```
Executes the test suite using the Node.js test runner.

Technical Details

SDK: Built with @lmstudio/sdk v1.5.0.
Language: TypeScript.
Search Engine: Utilizes DuckDuckGo's HTML interface with randomized User-Agents and rate limiting (2 seconds between requests) to ensure reliability.

License

Stats

33 Downloads

Stats

33 Downloads

Last updated

Project Files

src

test

.gitignore

31 B

LICENSE

1 KB

manifest.json

110 B

package-lock.json

29 KB

package.json

754 B

README.md

3 KB

tsconfig.json

359 B

README

LM Studio Web Search Plugin

A TypeScript-based web search plugin for LM Studio. This plugin enables local LLMs to access the live web, read page content, and perform pattern matching within web pages.

Features

Live Web Search: Query the web using DuckDuckGo (no API key required).
Page Reading: Extract clean, readable text from any URL.
In-Page Search: Find specific text or regex-style patterns within a web page.
Domain Filtering: Restrict searches and page reads to specific allowed domains.
Configurable Context: Adjust the amount of data returned to the model (low, medium, or high).
Today's Date: Get the current local date and time.
Privacy Focused: Operates locally as an LM Studio plugin.

Tools Included

The plugin registers four main tools with LM Studio:

search: Performs web searches. Supports single queries or batches of up to 5 queries.
read: Opens a URL and extracts its text content, automatically stripping HTML tags, scripts, and styles.
find_in_page: Searches for a specific pattern (string or regex) within a given URL and returns the surrounding context.
today: Returns the current local date and time.

Configuration

You can configure the following settings in LM Studio:

Search Context Size: Controls the default search breadth and page extraction size (low, medium, high).
Allowed Domains: An optional allowlist of domains for search results and page fetches.
Location Settings: Set your country, region, city, and timezone to improve search relevance.
Search Results Per Page: Customize the number of results returned (1-10).
Safe Search: Set the safety level for search results (strict, moderate, off).

Installation & Development

Prerequisites

Node.js (v18 or later recommended)
LM Studio installed

Setup

Clone this repository to your local machine.
Install dependencies:
```
npm install
```

Development Commands

Run in Development Mode:
```
npm run dev
```
This uses lms dev to load the plugin into LM Studio for testing.
Build the Plugin:
```
npm run build
```
Compiles the TypeScript source code into the dist directory.
Run Tests:
```
npm test
```
Executes the test suite using the Node.js test runner.

Technical Details

SDK: Built with @lmstudio/sdk v1.5.0.
Language: TypeScript.
Search Engine: Utilizes DuckDuckGo's HTML interface with randomized User-Agents and rate limiting (2 seconds between requests) to ensure reliability.

License

ugaldebarrio/web-search • LM Studio Hub

Stats

33 Downloads

Stats

33 Downloads

Last updated

Project Files

src

test

.gitignore

31 B

LICENSE

1 KB

manifest.json

110 B

package-lock.json

29 KB

package.json

754 B

README.md

3 KB

tsconfig.json

359 B

README

LM Studio Web Search Plugin

A TypeScript-based web search plugin for LM Studio. This plugin enables local LLMs to access the live web, read page content, and perform pattern matching within web pages.

Features

Live Web Search: Query the web using DuckDuckGo (no API key required).
Page Reading: Extract clean, readable text from any URL.
In-Page Search: Find specific text or regex-style patterns within a web page.
Domain Filtering: Restrict searches and page reads to specific allowed domains.
Configurable Context: Adjust the amount of data returned to the model (low, medium, or high).
Today's Date: Get the current local date and time.
Privacy Focused: Operates locally as an LM Studio plugin.

Tools Included

The plugin registers four main tools with LM Studio:

search: Performs web searches. Supports single queries or batches of up to 5 queries.
read: Opens a URL and extracts its text content, automatically stripping HTML tags, scripts, and styles.
find_in_page: Searches for a specific pattern (string or regex) within a given URL and returns the surrounding context.
today: Returns the current local date and time.

Configuration

You can configure the following settings in LM Studio:

Search Context Size: Controls the default search breadth and page extraction size (low, medium, high).
Allowed Domains: An optional allowlist of domains for search results and page fetches.
Location Settings: Set your country, region, city, and timezone to improve search relevance.
Search Results Per Page: Customize the number of results returned (1-10).
Safe Search: Set the safety level for search results (strict, moderate, off).

Installation & Development

Prerequisites

Node.js (v18 or later recommended)
LM Studio installed

Setup

Clone this repository to your local machine.
Install dependencies:
```
npm install
```

Development Commands

Run in Development Mode:
```
npm run dev
```
This uses lms dev to load the plugin into LM Studio for testing.
Build the Plugin:
```
npm run build
```
Compiles the TypeScript source code into the dist directory.
Run Tests:
```
npm test
```
Executes the test suite using the Node.js test runner.

Technical Details

SDK: Built with @lmstudio/sdk v1.5.0.
Language: TypeScript.
Search Engine: Utilizes DuckDuckGo's HTML interface with randomized User-Agents and rate limiting (2 seconds between requests) to ensure reliability.

License

Stats

33 Downloads

Stats

33 Downloads

Last updated

Project Files

src

test

.gitignore

31 B

LICENSE

1 KB

manifest.json

110 B

package-lock.json

29 KB

package.json

754 B

README.md

3 KB

tsconfig.json

359 B

README

LM Studio Web Search Plugin

A TypeScript-based web search plugin for LM Studio. This plugin enables local LLMs to access the live web, read page content, and perform pattern matching within web pages.

Features

Live Web Search: Query the web using DuckDuckGo (no API key required).
Page Reading: Extract clean, readable text from any URL.
In-Page Search: Find specific text or regex-style patterns within a web page.
Domain Filtering: Restrict searches and page reads to specific allowed domains.
Configurable Context: Adjust the amount of data returned to the model (low, medium, or high).
Today's Date: Get the current local date and time.
Privacy Focused: Operates locally as an LM Studio plugin.

Tools Included

The plugin registers four main tools with LM Studio:

search: Performs web searches. Supports single queries or batches of up to 5 queries.
read: Opens a URL and extracts its text content, automatically stripping HTML tags, scripts, and styles.
find_in_page: Searches for a specific pattern (string or regex) within a given URL and returns the surrounding context.
today: Returns the current local date and time.

Configuration

You can configure the following settings in LM Studio:

Search Context Size: Controls the default search breadth and page extraction size (low, medium, high).
Allowed Domains: An optional allowlist of domains for search results and page fetches.
Location Settings: Set your country, region, city, and timezone to improve search relevance.
Search Results Per Page: Customize the number of results returned (1-10).
Safe Search: Set the safety level for search results (strict, moderate, off).

Installation & Development

Prerequisites

Node.js (v18 or later recommended)
LM Studio installed

Setup

Clone this repository to your local machine.
Install dependencies:
```
npm install
```

Development Commands

Run in Development Mode:
```
npm run dev
```
This uses lms dev to load the plugin into LM Studio for testing.
Build the Plugin:
```
npm run build
```
Compiles the TypeScript source code into the dist directory.
Run Tests:
```
npm test
```
Executes the test suite using the Node.js test runner.

Technical Details

SDK: Built with @lmstudio/sdk v1.5.0.
Language: TypeScript.
Search Engine: Utilizes DuckDuckGo's HTML interface with randomized User-Agents and rate limiting (2 seconds between requests) to ensure reliability.

License