MCP Integration
Connect your Castiz video library to AI assistants using the Model Context Protocol (MCP). Enable semantic video search directly from Cursor, Windsurf, Claude Desktop, and more.
What is MCP?
The Model Context Protocol (MCP) is an open standard that allows AI assistants to connect with external data sources and tools. Castiz provides an MCP server that exposes your video library to AI assistants, enabling natural language video search.
Semantic Search
Search videos by meaning, not just keywords. Works in English and Hebrew.
Instant Results
AI-powered embeddings provide fast, relevant results from your library.
AI-Native
Works seamlessly with AI coding assistants and chat interfaces.
Supported Platforms
Cursor
AI-powered code editor
~/.cursor/mcp.jsonWindsurf
AI development environment
~/.windsurf/mcp.jsonClaude Desktop
Anthropic's Claude assistant
~/Library/Application Support/Claude/claude_desktop_config.jsonQuick Start
1. Get your API Key
Go to your Dashboard → Settings → API Keys and create a new API key with read scope.
2. Configure your AI assistant
Add the Castiz MCP server to your configuration file. The location depends on your platform:
~/.cursor/mcp.json~/.windsurf/mcp.json~/Library/Application Support/Claude/claude_desktop_config.json{
"mcpServers": {
"castiz": {
"command": "uvx",
"args": ["--from", "castiz-mcp", "castiz-mcp"],
"env": {
"CASTIZ_API_KEY": "sk_live_your_api_key_here"
}
}
}
}Alternative using npx:
{
"mcpServers": {
"castiz": {
"command": "npx",
"args": ["-y", "castiz-mcp"],
"env": {
"CASTIZ_API_KEY": "sk_live_your_api_key_here"
}
}
}
}3. Restart your AI assistant
Quit and reopen your AI assistant. The Castiz tools should now be available.
4. Try it out
Ask your AI assistant to search your videos:
"Search for holiday videos in my Castiz library"
"Find videos about cooking tutorials"
"What videos do I have about product demos?"
Available Tools
search_videos
Search for videos using natural language queries
Parameters
query*Natural language search query (supports Hebrew and English)
limitMaximum number of results (default: 20, max: 100)
similarity_thresholdMinimum similarity score 0-1 (default: 0.5)
Response Format
The search tool returns a JSON object with the matching videos and query interpretation:
{
"videos": [
{
"id": "181e00e6-496c-4821-9d6b-7451734318ed",
"uri": "/videos/181e00e6-496c-4821-9d6b-7451734318ed",
"name": "Holiday Songs Collection",
"description": "A collection of holiday songs for children...",
"thumbnail": "https://cdn.castiz.com/thumbnails/.../thumb.jpg",
"duration": 102,
"tags": ["holiday", "songs", "children"],
"relevance_score": 0.87,
"match_type": "semantic"
}
],
"total_count": 1,
"query_interpretation": {
"original_query": "holiday songs for kids",
"normalized_query": "holiday songs for kids",
"detected_tags": [],
"semantic_terms": ["holiday", "songs", "kids"],
"search_strategy": "semantic"
}
}Response Fields
| Field | Description |
|---|---|
| id | Unique video identifier (UUID) |
| uri | API path to the video resource |
| name | Video title |
| description | Video description |
| thumbnail | URL to thumbnail image |
| duration | Duration in seconds |
| tags | Array of tag names |
| relevance_score | Similarity score (0-1) |
| match_type | How matched: semantic, tag, or hybrid |
Troubleshooting
Tools not showing up
Make sure you've restarted your AI assistant after adding the configuration. Check that the config file path is correct for your platform.
Authentication errors
Verify your API key is correct and has not been revoked. API keys start with sk_live_ or sk_test_.
No results returned
Try lowering the similarity_threshold parameter (e.g., 0.3). Make sure your videos have been indexed for semantic search.
REST API Alternative
You can also use the search functionality via REST API:
curl -X POST "https://api.castiz.com/skills/search_videos/execute" \
-H "X-API-Key: sk_live_your_api_key_here" \
-H "Content-Type: application/json" \
-d '{"query": "holiday songs", "limit": 10}'