This document explains how to set up the HackerNews cron job on Val.Town.

The hackernews-cron.ts file contains a cron job that:

• Runs every hour
• Fetches the top 10 HackerNews stories
• Automatically generates videos for new URLs (no duplicates)
• Uses your default video generation prompt
• Skips text posts (Ask HN, Show HN, etc.) - only processes external URLs

1. Go to Val.Town
2. Create a new Val
3. Copy the contents of hackernews-cron.ts into the Val
4. Save the Val

The cron job directly imports and uses your existing video generation logic, so it requires the same environment variables as your main application:

Required:

• GEMINI_API_KEY - Your Google Gemini API key (required for video generation)

Optional:

• ADMIN_PASSWORD - Your admin password (if you want to use the admin interface)

Note: No additional environment variables are needed! The cron job directly calls the generateVideoAsync function instead of making HTTP requests, making it simpler and more efficient.

In Val.Town:

1. Click on the "Schedule" button in your Val
2. Set the schedule to run every hour: 0 * * * * (cron syntax)
3. Save the schedule

You can easily adjust the number of stories to process by editing the constant at the top of the file:

const TOP_STORIES_LIMIT = 10; // Change this to process more or fewer stories

Common values:

• 10 - Top 10 stories (default, recommended for cost control)
• 20 - Top 20 stories
• 50 - Top 50 stories
• 100 - Top 100 stories

1. Fetch Stories: The cron job fetches the list of top story IDs from HackerNews
2. Get Story Details: For each story (up to TOP_STORIES_LIMIT), it fetches the full story details
3. Filter: It filters out:
   • Stories without external URLs (text posts, Ask HN, Show HN)
   • URLs that already exist in the database (duplicates)
4. Generate Videos: For each new URL, it directly calls the generateVideoAsync() function with the default prompt
5. Log Results: It logs a summary of processed, skipped, and failed stories

The cron job uses the getVideoByUrl() database query to check if a URL has already been processed. This ensures:

• Cost control: You don't regenerate videos for the same URL
• Efficiency: No wasted API calls or processing time
• Data integrity: Each URL is only processed once

Note: If you want to regenerate a video for a URL, you can do so through the admin interface.

The cron job only processes stories with external URLs. It skips:

• Ask HN posts (text-only discussions)
• Show HN posts without URLs
• Job postings without URLs
• Any other text-only content

This ensures you only generate videos for stories that have actual web content to summarize.

The cron job logs detailed information about its execution:

[HN Cron] Starting HackerNews processing at 2025-10-21T22:00:00.000Z
[HN Cron] Processing top 10 stories
[HN Cron] Retrieved 500 total story IDs
[HN Cron] Processing 10 stories
[HN Cron] Story 12345: Generating video for https://example.com
[HN Cron] Story 12345: Video generation started successfully
[HN Cron] Story 12346: URL already processed, skipping
[HN Cron] Story 12347: No external URL, skipping
[HN Cron] Processing complete:
[HN Cron] - Processed: 3
[HN Cron] - Skipped: 7
[HN Cron] - Errors: 0

You can view these logs in Val.Town's execution logs.

Before setting up the cron schedule, you can test the job manually:

1. In Val.Town, open your cron job Val
2. Click the "Run" button to execute it immediately
3. Check the logs to see the results
4. Verify that videos are being generated in your application

Possible causes:

1. All URLs might already be in the database (check logs for "duplicate" messages)
2. All stories might be text posts without URLs (check logs for "no URL" messages)
3. The GEMINI_API_KEY might be missing or invalid

Solution: Check the execution logs for specific error messages.

Solution: Reduce the TOP_STORIES_LIMIT constant to a lower number (e.g., 5 instead of 10).

The cron job is designed with cost control in mind:

1. Configurable limit: Easy to adjust the number of stories processed
2. Duplicate prevention: Never processes the same URL twice
3. Efficient filtering: Skips text posts before making API calls
4. Batched processing: All video generations are triggered at once, then processed asynchronously

To further control costs:

• Set TOP_STORIES_LIMIT to a lower number
• Run the cron less frequently (e.g., every 2-3 hours instead of hourly)
• Monitor the logs to see how many new stories are typically found

The cron job integrates seamlessly with your existing system:

• Uses the same video generation endpoint (/api/videos/generate)
• Leverages the existing database schema and queries
• Works with the existing polling mechanism for video completion
• Respects the default video prompt from your configuration
• Videos can be managed through the admin interface like any other video

No changes to your existing code are required - the cron job is completely standalone.