searchDocs

Search the Fluvie documentation using TF-IDF (Term Frequency-Inverse Document Frequency) ranking.

Overview

The searchDocs tool provides intelligent documentation search with:

• TF-IDF ranking for relevance scoring
• Category filtering to narrow results
• Snippet extraction with context
• Configurable result limits

Input Schema

{
  "type": "object",
  "properties": {
    "query": {
      "type": "string",
      "description": "Search query text"
    },
    "category": {
      "type": "string",
      "description": "Filter by documentation category",
      "enum": ["widgets", "animations", "effects", "templates", "advanced", "getting-started"]
    },
    "limit": {
      "type": "number",
      "description": "Maximum number of results (1-10)",
      "minimum": 1,
      "maximum": 10,
      "default": 5
    }
  },
  "required": ["query"]
}

Examples

Basic Search

{
  "query": "how to fade text"
}

Category-Filtered Search

{
  "query": "layer animations",
  "category": "widgets"
}

Limited Results

{
  "query": "ffmpeg configuration",
  "limit": 3
}

Response Format

{
  "results": [
    {
      "title": "Document Title",
      "path": "relative/path/to/doc.md",
      "category": "widgets",
      "snippet": "...relevant text excerpt with search terms...",
      "score": 0.85
    }
  ],
  "total": 1,
  "query": "original query"
}

How TF-IDF Works

The search engine uses TF-IDF to rank documents:

1. Term Frequency (TF): How often does the search term appear in this document?
2. Inverse Document Frequency (IDF): How unique is this term across all documents?
3. Score: TF × IDF gives higher scores to documents where search terms are both frequent AND distinctive

This means:

• Common words like “the” have low IDF (appear everywhere)
• Specific terms like “VideoComposition” have high IDF (appear in few documents)
• A document mentioning “VideoComposition” many times scores highest

Tips for Effective Searches

1. Use specific terms: “LayerStack fadeIn” is better than “how to fade”
2. Include widget names: “AnimatedText typewriter” finds relevant docs faster
3. Use category filter: Narrow results when you know the topic area
4. Combine concepts: “audio sync beat detection” finds intersection of topics