diff --git a/.env.example b/.env.example
index cd6c51f..2069c94 100644
--- a/.env.example
+++ b/.env.example
@@ -18,6 +18,10 @@ ANTHROPIC_MODEL_CHAT=claude-3-5-sonnet-latest
 # --- Google Gemini (optional, used by MCP server) ---
 GOOGLE_API_KEY=
 
+# --- Search Provider ---
+SEARCH_PROVIDER=duckduckgo     # duckduckgo | tavily
+TAVILY_API_KEY=                # Required when SEARCH_PROVIDER=tavily
+
 # --- Storage ---
 CHROMA_DIR=.chroma
 SQLITE_PATH=.sqlite/agent.db
diff --git a/mcp/server.py b/mcp/server.py
index 87a4bf8..4279828 100644
--- a/mcp/server.py
+++ b/mcp/server.py
@@ -71,8 +71,8 @@ async def summarize(req: SummarizeRequest) -> Dict[str, Any]:
 
         @self.app.get("/search")
         async def search(q: str, max_results: int = 5) -> Dict[str, Any]:
-            """Perform a web search using DuckDuckGo."""
-            results = await webtools.search_ddg(q, max_results=max_results)
+            """Perform a web search."""
+            results = await webtools.search_web(q, max_results=max_results)
             return {"query": q, "results": results}
 
         @self.app.get("/browse")
@@ -84,7 +84,7 @@ async def browse(url: str) -> Dict[str, Any]:
         @self.app.get("/research")
         async def research(q: str, max_results: int = 3) -> Dict[str, Any]:
             """Conduct a search and fetch the contents of top results."""
-            results = await webtools.search_ddg(q, max_results=max_results)
+            results = await webtools.search_web(q, max_results=max_results)
             pages: List[Dict[str, str]] = []
             for res in results:
                 url = res.get("href") or res.get("url")
diff --git a/mcp/tools/web.py b/mcp/tools/web.py
index 5f3ba36..9d6dc3d 100644
--- a/mcp/tools/web.py
+++ b/mcp/tools/web.py
@@ -1,5 +1,7 @@
 from __future__ import annotations
 
+import os
+
 import httpx
 import trafilatura
 from bs4 import BeautifulSoup
@@ -14,6 +16,33 @@ async def search_ddg(q: str, max_results: int = 5):
             return []
 
 
+async def search_tavily(q: str, max_results: int = 5):
+    """Search using the Tavily API, returning results normalised to {title, href, body}."""
+    from tavily import AsyncTavilyClient
+
+    try:
+        client = AsyncTavilyClient(api_key=os.environ.get("TAVILY_API_KEY", ""))
+        response = await client.search(query=q, max_results=max_results)
+    except Exception:
+        return []
+    results = []
+    for r in response.get("results", []):
+        results.append({
+            "title": r.get("title", ""),
+            "href": r.get("url", ""),
+            "body": r.get("content", ""),
+        })
+    return results
+
+
+async def search_web(q: str, max_results: int = 5):
+    """Dispatch to the configured search provider (duckduckgo or tavily)."""
+    provider = os.environ.get("SEARCH_PROVIDER", "duckduckgo").lower()
+    if provider == "tavily":
+        return await search_tavily(q, max_results=max_results)
+    return await search_ddg(q, max_results=max_results)
+
+
 async def fetch_page(url: str) -> str:
     async with httpx.AsyncClient() as client:
         resp = await client.get(url, timeout=15)
diff --git a/requirements.txt b/requirements.txt
index d10d119..2c0f0c6 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -17,6 +17,7 @@ typing-extensions>=4.12.2
 
 # Discovery
 duckduckgo-search>=6.2.10
+tavily-python>=0.3.0
 httpx>=0.27.0
 trafilatura>=1.10.0
 beautifulsoup4>=4.12.3