URL Retriever
The URL Retriever lets you fetch and process content from web links, supporting OCR, metadata extraction, and flexible output for powering AI workflows.
Component description
How the URL Retriever component works
URL Retriever Component
The URL Retriever is a versatile flow component designed to fetch and process web content from specified URLs, returning the information as structured documents. It serves as a bridge between external online content and your AI workflow, enabling you to integrate, analyze, or process web-based information efficiently.
What Does It Do?
This component retrieves the content of one or multiple URLs provided as input. It can extract the main text, metadata, and even process content from images using Optical Character Recognition (OCR). The retrieved data is then made available in various structured formats suitable for downstream AI tasks such as summarization, question answering, or knowledge extraction.
Input Options
You can supply URLs to the component in two ways:
Text URLs:
- Input Type:
Message - Description: A list of plain URL links for the component to fetch content from.
- Input Type:
URL Records:
- Input Type:
UrlRecord - Description: A list of structured URL records, which may include additional metadata.
- Input Type:
Advanced Input Parameters
| Parameter | Type | Default | Description |
|---|---|---|---|
| Apply OCR | Boolean | false | If enabled, applies OCR to extract text from images in the document. |
| Cache TTL | Dropdown | 2 weeks | How long the content should be cached, with options from no cache up to 1 year. |
| From H1 if exists | Boolean | true | Begins extraction from the H1 tag if present, focusing on main content. |
| Load from pointer | Boolean | true | Loads content starting from the most relevant section based on your query. |
| Hide Resources | Boolean | false | Hides the retrieved resources from being output or displayed. |
| Max Tokens | Integer | 3000 | Sets the maximum number of tokens for the output text. |
| Skip Last Header | Boolean | true | Skips the last header during extraction for streamlined content. |
| Strategy | Dropdown | Include equal size from each documents | Determines how content is combined: concatenate fully or include equal parts from each document. |
| Export Content | Multi-select | All | Choose which HTML elements to export (H1-H6, Paragraph). |
| Include Metadata | Multi-select | Product | Specify which metadata fields to include (e.g., Product, Author, Website, etc.). |
| Verbose | Boolean | false | Enables detailed output for debugging or information purposes. |
| Tool Name | String | (empty) | Optionally assign a custom name to the tool for agent reference. |
| Tool Description | Multiline | (empty) | Provide a description to help agents understand the tool’s purpose. |
Outputs
The URL Retriever provides its outputs in several formats, allowing flexible integration with various AI processes:
| Output Name | Type | Description |
|---|---|---|
| Documents | Message | The processed content from the URLs, ready for use in messaging-oriented workflows. |
| Raw Documents | Document | The raw, unprocessed document objects for advanced downstream processing. |
| Documents As Tool | Tool | The content packaged as a tool, enabling agent-based workflows to utilize the documents. |
Why Use the URL Retriever?
- Integrate External Knowledge: Seamlessly bring web-based information into your AI applications, such as chatbots, search engines, or knowledge bases.
- Customizable Extraction: Fine-tune what content and metadata you want, control the amount of data, and use OCR for images.
- Performance & Efficiency: Use caching to avoid redundant downloads, and limit token output for performance.
- Flexible Output Formats: Choose the output format that best fits your next workflow step—structured document, message, or tool.
Example Use Cases
- Building knowledge-grounded conversational agents that answer questions using up-to-date web content.
- Aggregating product data from e-commerce sites for comparison or analytics.
- Monitoring and analyzing blog or news articles based on specific topics or keywords.
- Extracting information from web pages containing mixed media (text and images).
Summary Table
| Feature | Description |
|---|---|
| Fetches URLs | Retrieves and processes web content from provided URLs. |
| OCR Support | Extracts text from images in documents if enabled. |
| Metadata Extraction | Optionally includes metadata such as author, product, or schema.org types. |
| Customizable Output | Select which HTML elements or metadata to export. |
| Caching | Configurable cache lifetimes for efficiency. |
| Multiple Output Types | Supports message, raw document, and tool outputs for workflow flexibility. |
The URL Retriever is a powerful and flexible bridge between web content and your AI workflows, offering granular control over content extraction and integration.
Examples of flow templates using URL Retriever component
To help you get started quickly, we have prepared several example flow templates that demonstrate how to use the URL Retriever component effectively. These templates showcase different use cases and best practices, making it easier for you to understand and implement the component in your own projects.
Frequently asked questions
- What does the URL Retriever component do?
The URL Retriever fetches and processes content from specified web links, making text and metadata from online documents available for your workflow or AI agent.
- Can it extract content from images or PDFs?
Yes, by enabling the OCR option, the component can extract text from image-based documents or scanned PDFs.
- What types of outputs does it provide?
It outputs processed documents as text messages, raw document objects, or as a tool for agent workflows, depending on your setup.
- How does caching work in URL Retriever?
You can set how long retrieved content is cached, reducing repeated downloads and speeding up your flows.
- Can I control what parts of a webpage are extracted?
Yes, you can specify which headings, paragraphs, or metadata fields to include in the output, allowing for targeted extraction.
- Is this suitable for building knowledge bots or web data automations?
Absolutely. The URL Retriever is essential for any automation or chatbot that needs to read, process, or summarize live web content.
Try FlowHunt URL Retriever
Supercharge your workflows by integrating live web content. Extract, process, and utilize data from URLs with ease.
Learn more
