feat: add Tavily as a pluggable internet search backend

examples/basic_modules/textual_memory_internet_search_example.py

@@ -288,6 +288,83 @@
     print("\n   Get your credentials from:")
     print("   https://developers.google.com/custom-search/v1/overview")
 
+# ============================================================================
+# Step 7: Test Tavily Search API (Optional)
+# ============================================================================
+print("\n" + "=" * 80)
+print("TAVILY SEARCH API TEST")
+print("=" * 80)
+
+tavily_api_key = os.environ.get("TAVILY_API_KEY", "")
+
+if tavily_api_key:
+    print("\n[Step 7.1] Configuring Tavily Search retriever...")
+
+    tavily_retriever_config = InternetRetrieverConfigFactory.model_validate(
+        {
+            "backend": "tavily",
+            "config": {
+                "api_key": tavily_api_key,
+                "max_results": 5,
+            },
+        }
+    )
+
+    print("✓ Tavily retriever configured")
+    print(f"  Max results: {tavily_retriever_config.config.max_results}")
+
+    print("\n[Step 7.2] Creating Tavily retriever instance...")
+    tavily_retriever = InternetRetrieverFactory.from_config(tavily_retriever_config, embedder)
+    print("✓ Tavily retriever initialized")
+
+    print("\n[Step 7.3] Performing Tavily web search...")
+    tavily_query = "latest AI research breakthroughs 2024"
+    print(f"  Query: '{tavily_query}'")
+    print("  Searching via Tavily Search API...\n")
+
+    tavily_results = tavily_retriever.retrieve_from_internet(tavily_query)
+
+    print("✓ Tavily search completed!")
+    print(f"✓ Retrieved {len(tavily_results)} memory items from Tavily search\n")
+
+    print("=" * 80)
+    print("TAVILY SEARCH RESULTS")
+    print("=" * 80)
+
+    if not tavily_results:
+        print("\nNo results found from Tavily.")
+        print("   This might indicate:")
+        print("   - Invalid Tavily API key")
+        print("   - API quota exceeded")
+        print("   - Network connectivity issues")
+    else:
+        for idx, item in enumerate(tavily_results, 1):
+            print(f"\n[Tavily Result #{idx}]")
+            print("-" * 80)
+
+            content = item.memory
+            if len(content) > 300:
+                print(f"Content: {content[:300]}...")
+                print(f"         (... {len(content) - 300} more characters)")
+            else:
+                print(f"Content: {content}")
+
+            if hasattr(item, "metadata") and item.metadata:
+                metadata = item.metadata
+                if hasattr(metadata, "sources") and metadata.sources:
+                    print(f"Source: {metadata.sources[0] if metadata.sources else 'N/A'}")
+
+            print()
+
+    print("=" * 80)
+    print("Tavily Search Test completed!")
+    print("=" * 80)
+else:
+    print("\n  Skipping Tavily Search API test")
+    print("   To enable this test, set the following environment variable:")
+    print("   - TAVILY_API_KEY: Your Tavily API key")
+    print("\n   Get your API key from: https://app.tavily.com")
+
 print("\n" + "=" * 80)
 print("ALL TESTS COMPLETED")
 print("=" * 80)
@@ -297,6 +374,10 @@
     print("  ✓ Tested Google Custom Search API")
 else:
     print("  ⏭️  Skipped Google Custom Search API (credentials not set)")
+if tavily_api_key:
+    print("  Tested Tavily Search API")
+else:
+    print("  Skipped Tavily Search API (credentials not set)")
 print("\n💡 Quick Start:")
 print("  # Set BochaAI API key")
 print("  export BOCHA_API_KEY='sk-your-bocha-api-key'")
@@ -305,5 +386,8 @@
 print("  export GOOGLE_API_KEY='your-google-api-key'")
 print("  export GOOGLE_SEARCH_ENGINE_ID='your-search-engine-id'")
 print("  ")
+print("  # Set Tavily API key (optional)")
+print("  export TAVILY_API_KEY='tvly-your-tavily-api-key'")
+print("  ")
 print("  # Run the example")
 print("  python examples/basic_modules/textual_memory_internet_search_example.py\n")

pyproject.toml

@@ -102,6 +102,11 @@ skill-mem = [
     "alibabacloud-oss-v2 (>=1.2.2,<1.2.3)",
 ]
 
+# Tavily Search
+tavily = [
+    "tavily-python (>=0.5.0)",
+]
+
 # All optional dependencies
 # Allow users to install with `pip install MemoryOS[all]`
 all = [
@@ -129,6 +134,7 @@ all = [
     "nltk (>=3.9.1,<4.0.0)",
     "rake-nltk (>=1.0.6,<1.1.0)",
     "alibabacloud-oss-v2 (>=1.2.2,<1.2.3)",
+    "tavily-python (>=0.5.0)",
 
     # Uncategorized dependencies
 ]

src/memos/api/config.py

@@ -626,7 +626,20 @@ def get_oss_config() -> dict[str, Any] | None:
         return config
 
     def get_internet_config() -> dict[str, Any]:
-        """Get embedder configuration."""
+        """Get internet retriever configuration."""
+        tavily_api_key = os.getenv("TAVILY_API_KEY", "")
+        bocha_api_key = os.getenv("BOCHA_API_KEY", "")
+
+        # Use Tavily if TAVILY_API_KEY is set and BOCHA_API_KEY is absent
+        if tavily_api_key and not bocha_api_key:
+            return {
+                "backend": "tavily",
+                "config": {
+                    "api_key": tavily_api_key,
+                    "max_results": 15,
+                },
+            }
+
         reader_config = APIConfig.get_reader_config()
         return {
             "backend": "bocha",

src/memos/configs/internet_retriever.py

@@ -67,6 +67,12 @@ class BochaSearchConfig(BaseInternetRetrieverConfig):
     )
 
 
+class TavilySearchConfig(BaseInternetRetrieverConfig):
+    """Configuration class for Tavily Search API."""
+
+    max_results: int = Field(default=20, description="Maximum number of results to retrieve")
+
+
 class InternetRetrieverConfigFactory(BaseConfig):
     """Factory class for creating internet retriever configurations."""
 
@@ -82,6 +88,7 @@ class InternetRetrieverConfigFactory(BaseConfig):
         "bing": BingSearchConfig,
         "xinyu": XinyuSearchConfig,
         "bocha": BochaSearchConfig,
+        "tavily": TavilySearchConfig,
     }
 
     @field_validator("backend")

src/memos/memories/textual/tree_text_memory/retrieve/internet_retriever_factory.py

@@ -9,6 +9,7 @@
 from memos.memories.textual.tree_text_memory.retrieve.internet_retriever import (
     InternetGoogleRetriever,
 )
+from memos.memories.textual.tree_text_memory.retrieve.tavilysearch import TavilySearchRetriever
 from memos.memories.textual.tree_text_memory.retrieve.xinyusearch import XinyuSearchRetriever
 from memos.memos_tools.singleton import singleton_factory
 
@@ -21,6 +22,7 @@ class InternetRetrieverFactory:
         "bing": InternetGoogleRetriever,  # TODO: Implement BingRetriever
         "xinyu": XinyuSearchRetriever,
         "bocha": BochaAISearchRetriever,
+        "tavily": TavilySearchRetriever,
     }
 
     @classmethod
@@ -81,6 +83,12 @@ def from_config(
                 reader=MemReaderFactory.from_config(config.reader),
                 max_results=config.max_results,
             )
+        elif backend == "tavily":
+            return retriever_class(
+                api_key=config.api_key,
+                embedder=embedder,
+                max_results=config.max_results,
+            )
         else:
             raise ValueError(f"Unsupported backend: {backend}")