A powerful search engine for Val Town projects that enables searching across code files, project descriptions, and user data.

• Full-text search across all project files and content
• Enhanced project descriptions generated using AI analysis of README files and code
• Multi-type search supporting files, projects, users, and documentation
• Real-time typeahead suggestions for usernames and project names
• Newsletter curation tool for finding active projects
• JSON API access for programmatic usage

The system now automatically generates concise, one-line descriptions for projects by analyzing:

1. Existing project descriptions from Val Town API
2. README files (README.md, README.txt, etc.)
3. Main code files (index.ts, main.js, app.tsx, etc.)
4. File structure and naming for additional context

When importing or updating projects, the system:

1. Checks if the project has a good existing description
2. Looks for README files in the project root
3. Analyzes main code files if no README is found
4. Uses OpenAI to generate a concise summary (under 100 characters)
5. Stores the enhanced description in the database

You can manually trigger description enhancement for existing projects:

GET /enhance-descriptions?limit=50

This endpoint will:

• Find projects with poor or missing descriptions
• Generate AI summaries for up to the specified limit
• Return statistics about the enhancement process

Projects are considered to need enhanced descriptions if they:

• Have no description
• Have empty descriptions
• Have descriptions shorter than 10 characters
• Contain placeholder text like "todo" or "description"

• GET /?q=searchterm - Main search interface
• GET /?q=searchterm&format=json - JSON API access
• GET /?q=searchterm&type=files|projects|users|docs - Filter by result type

• GET /typeahead?q=partial - Get typeahead suggestions

• GET /newsletter?days=30 - Find active projects for newsletter
• GET /newsletter?days=30&format=json - JSON format

• GET /enhance-descriptions?limit=10 - Enhance project descriptions

The AI generates concise, action-oriented descriptions:

• "Creates interactive data visualizations from CSV files"
• "Provides real-time weather data API with caching"
• "Generates QR codes with custom styling options"
• "Converts Markdown to HTML with syntax highlighting"

The system requires:

• SQLite database for storing project and file data
• OpenAI API access for description generation
• Val Town SDK for data import

Environment variables:

• OpenAI API key (automatically available in Val Town environment)

The system uses three main tables:

• projects - Project metadata and enhanced descriptions
• files - File content for search indexing
• users - User profile information

To contribute to the project description enhancement:

1. Modify the prompt in enhance-descriptions.ts to improve summary quality
2. Adjust the enhancement criteria in db.ts
3. Add new file type support for analysis
4. Improve the README detection logic

• Description enhancement processes 10-50 projects per request
• Includes rate limiting to avoid API throttling
• Uses efficient database queries with pagination
• Caches user data to avoid redundant API calls