LiteLLM
Runtime provider id: litellm. Routes search through a LiteLLM proxy’s /v1/search/<tool> endpoint — useful for centralizing search credentials/routing behind a gateway you already run for models.
Options
Section titled “Options”| Field | Type | Notes |
|---|---|---|
baseURL |
string | Required. Base URL of the LiteLLM proxy. |
apiKey |
string | Required. Bearer token for the proxy. |
tool |
string | Required. The underlying search tool name the proxy should invoke (e.g. "tavily_search"). |
Supported search parameters
Section titled “Supported search parameters”LiteLLM supports count (max_results) and country. freshness, start_date/end_date, and safe_search are not forwarded. Requests use the longer 60-second timeout shared with firecrawl.
Example config
Section titled “Example config”{ "search": { "enabled": true, "provider": "litellm", "litellm": { "baseURL": "https://litellm.internal", "apiKey": "sk-litellm-...", "tool": "tavily_search" } }}