Tutorial: Building a HackerNews Filter Flow

In this tutorial, we'll create a flow that scrapes HackerNews and uses an LLM to filter posts based on topics you're interested in. This flow will help you quickly find articles about topics you care about.

Overview of What We're Building

This flow will:

1. Scrape content from HackerNews (from either the front page or newest posts)
2. Use an LLM to filter and extract articles matching your topic of interest
3. Return the relevant articles as clickable links

Step 1: Create a New Flow

Start by clicking "Create Flow" in your workspace's agent skills page.

Step 2: Configure Flow Information

In the Flow Information block, set up:

Name:

Hacker News Headline Viewer

Description:

This tool can be used to visit hacker news webpage and extract ALL headlines and links from the page that have to do with a particular topic.

Available options for `page`:
(empty) - front page
"newest" - newest posts page

Examples of how to use this flow:
"Find AI-related posts on HackerNews"
"Show me political discussions from the newest HackerNews posts"

The flow will return relevant articles as clickable markdown links.

Your flow info block should look like this:

Step 3: Set Up Flow Variables

In the Flow Variables block, create these variables:

1. hackerNewsURLPath

Name: hackerNewsURLPath
Default Value: (leave empty)

2. topicOfInterest

Name: topicOfInterest
Default Value: Political discussions or items

3. pageContentFromSite

Name: pageContentFromSite
Default Value: (leave empty)

Your flow start block should look like this:

Step 4: Add Web Scraping Block

1. Click "Add Block" below the Flow Variables block
2. Select "Web Scraper"
3. Configure the block:

URL to scrape: https://news.ycombinator.com/${hackerNewsURLPath}
Result Variable: pageContentFromSite

Your web scraping block should look like this:

Step 5: Add LLM Instruction Block

1. Click "Add Block" below the Web Scraper block
2. Select "LLM Instruction"
3. Configure the block:

Input Variable: pageContentFromSite

Instructions:

Extract all links from this content that would be relevant to this topic: ${topicOfInterest}

Content:
${pageContentFromSite}

Format your response as a list of markdown links, with a brief description of why each link is relevant.
If no relevant links are found, say "No relevant articles found."

Result Variable: (leave empty)

Your LLM instruction block should look like this:

Step 6: Save and Test

1. Click "Save" in the top right corner
2. Disable other agent skills to ensure this flow is used
3. Test the flow with prompts like:

Find AI-related posts on HackerNews

Show me political discussions from the newest HackerNews posts

What are the latest cryptocurrency articles on HackerNews?

Example output:

Customizing the Flow

You can enhance this flow by:

• Adding more specific filtering criteria in the LLM instructions
• Including additional variables for more complex filtering
• Modifying the output format to include more details about each article

Troubleshooting

If you're not getting the expected results:

1. Check that your flow variables are correctly named
2. Verify the Web Scraper URL is correct
3. Make sure your LLM instructions are clear and specific
4. Review the debugging flows guide for more help