Web Search
Perform web searches to find real-time information from the internet.
Parameters
| Name | Type | Required | Default | Description |
|---|---|---|---|---|
query | string | ✅ Yes | - | The search query. |
maxResults | number | ❌ No | 5 | Maximum results to return (1–20). |
searchDepth | enum (basic | advanced) | ❌ No | basic | Tavily search depth. advanced uses better extraction at 2× the upstream cost; recommended when includeRawContent is true. |
includeRawContent | boolean | ❌ No | false | When true, each result includes a rawContent field with the full cleaned page body. No extra credit cost; expect a small latency increase. |
Cost
| Mode | Credits per call |
|---|---|
searchDepth: "basic" (default) | 0.01 |
searchDepth: "advanced" | 0.02 |
includeRawContent is free at any search depth — it only adds latency. The
credit cost is reported per call via _meta.creditCost on the MCP response,
so dashboards and the agent-service trace ledger pick up the right amount
automatically.
Required Integrations
tavily
Use Cases
- General research
- Current events
- Supplemental data gathering
- Real-time information
Example Usage
MCP Client (default — basic, snippets only)
{
"tool": "web_search",
"parameters": {
"query": "example-value"
}
}
MCP Client (full content + advanced extraction)
{
"tool": "web_search",
"parameters": {
"query": "Phoenix MCP architecture changelog",
"searchDepth": "advanced",
"includeRawContent": true
}
}
This call costs 0.02 credits and returns each result with rawContent
populated for downstream RAG / extraction workflows.
Claude Desktop
Use the web_search tool to perform web searches to find real-time information from the internet.
Response Format
The tool returns a structured JSON response with:
data: The actual result datametadata: Execution metadata including timing and provider information
When includeRawContent: true, each entry in data.results carries a
rawContent string (otherwise null).
Error Handling
Common errors:
- Missing Integration: Required integration not configured
- Invalid Parameters: Parameters don't match the schema
- Rate Limit: API rate limit exceeded
- Not Found: Resource not found (e.g., company domain doesn't exist)
Related Tools
- See all available tools for more options