Update mcp-searxng instead of DuckDuckGoSearchTool

app.py

@@ -11,10 +11,11 @@ from mcp import StdioServerParameters
 from smolagents import DuckDuckGoSearchTool, FinalAnswerTool, Tool, ToolCollection, VisitWebpageTool
 from ycecream import y
 
-from basic_agent import BasicAgent
+from basic_agent import BasicAgent, WikipediaSearchTool
 from get_model import get_model
+from openai_model import openai_model
 
-y.configure(sln=1)
+y.configure(sln=0)
 print = rich.get_console().print
 
 # (Keep Constants as is)
@@ -55,24 +56,6 @@ def run_and_submit_all(profile: gr.OAuthProfile | None):
     questions_url = f"{api_url}/questions"  # https://agents-course-unit4-scoring.hf.space/questions
     submit_url = f"{api_url}/submit"
 
-    # 1. Instantiate Agent ( modify this part to create your agent)
-    try:
-        agent = BasicAgent(
-            # model=get_model(cat="gemini"),
-            model=get_model(cat="llama"),
-            tools=[
-                DuckDuckGoSearchTool(),
-                VisitWebpageTool(),
-                # FinalAnswerTool(),
-            ],
-        )
-    except Exception as e:
-        print(f"Error instantiating agent: {e}")
-        return f"Error initializing agent: {e}", None
-    # In the case of an app running as a hugging Face space, this link points toward your codebase ( usefull for others so please keep it public)
-    agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
-    print(agent_code)
-
     # 2. Fetch Questions
     print(f"Fetching questions from: {questions_url}")
     try:
@@ -95,27 +78,75 @@ def run_and_submit_all(profile: gr.OAuthProfile | None):
         print(f"An unexpected error occurred fetching questions: {e}")
         return f"An unexpected error occurred fetching questions: {e}", None
 
-    # 3. Run your Agent
-    results_log = []
-    answers_payload = []
-    print(f"Running agent on {len(questions_data)} questions...")
-    for item in questions_data:
-        task_id = item.get("task_id")
-        question_text = item.get("question")
-        if not task_id or question_text is None:
-            print(f"Skipping item with missing task_id or question: {item}")
-            continue
+    # Prepare model and mcp_params
+    model = openai_model()  # defautl llama4 scout
+
+    # messages = [{'role': 'user', 'content': 'Say this is a test.'}]
+    # print(model(messages))
+
+    # raise SystemExit("By intention")
+
+    mcp_searxng_params = StdioServerParameters(
+        **{
+          "command": "npx",
+          "args": [
+            "-y",
+            "mcp-searxng"
+          ],
+          "env": {
+            "SEARXNG_URL": "https://searx.dattw.eu.org"
+          }
+        }
+    )
+
+    # with ToolCollection.from_mcp(mcp_searxng_params, trust_remote_code=True) as searxng_tool_collection, ToolCollection.from_mcp(mcp_markitdown_params, trust_remote_code=True) as markitdown_tools:
+    with ToolCollection.from_mcp(mcp_searxng_params, trust_remote_code=True) as searxng_tool_collection:
+        # 1. Instantiate Agent ( modify this part to create your agent)
         try:
-            submitted_answer = agent(question_text)
-            answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
-            results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted_answer})
+            agent = BasicAgent(
+                # model=get_model(cat="gemini"),
+                # model=get_model(cat="llama"),
+                model=model,
+                tools=[
+                    *searxng_tool_collection.tools,
+                    # DuckDuckGoSearchTool(),
+                    VisitWebpageTool(),
+                    WikipediaSearchTool(),
+                    FinalAnswerTool(),
+                ],
+                verbosity_level=1,
+            )
+            agent.agent.visualize()
         except Exception as e:
-            print(f"Error running agent on task {task_id}: {e}")
-            results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": f"AGENT ERROR: {e}"})
-
-    if not answers_payload:
-        print("Agent did not produce any answers to submit.")
-        return "Agent did not produce any answers to submit.", pd.DataFrame(results_log)
+            print(f"Error instantiating agent: {e}")
+            return f"Error initializing agent: {e}", None
+        # In the case of an app running as a hugging Face space, this link points toward your codebase ( usefull for others so please keep it public)
+        agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
+        print(agent_code)
+
+        # 3. Run your Agent
+        results_log = []
+        answers_payload = []
+        print(f"Running agent on {len(questions_data)} questions...")
+        for item in questions_data:
+            task_id = item.get("task_id")
+            question_text = item.get("question")
+            if not task_id or question_text is None:
+                print(f"Skipping item with missing task_id or question: {item}")
+                continue
+            try:
+                submitted_answer = agent(question_text)
+                answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
+                results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted_answer})
+            except Exception as e:
+                print(f"Error running agent on task {task_id}: {e}")
+                results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": f"AGENT ERROR: {e}"})
+
+        if not answers_payload:
+            print("Agent did not produce any answers to submit.")
+            return "Agent did not produce any answers to submit.", pd.DataFrame(results_log)
+
+        agent.agent.visualize()
 
     # 4. Prepare Submission
     submission_data = {"username": username.strip(), "agent_code": agent_code, "answers": answers_payload}

basic_agent.py

@@ -11,7 +11,8 @@ import rich
 import smolagents
 import wikipediaapi
 from loguru import logger
-from smolagents import CodeAgent, DuckDuckGoSearchTool, FinalAnswerTool, Tool, VisitWebpageTool
+from mcp import StdioServerParameters
+from smolagents import CodeAgent, DuckDuckGoSearchTool, FinalAnswerTool, Tool, ToolCollection, VisitWebpageTool
 from smolagents import InferenceClientModel as HfApiModel
 
 from get_model import get_model
@@ -78,6 +79,8 @@ class BasicAgent:
     model: smolagents.models.Model = HfApiModel()
     tools: list = field(default_factory=lambda: [])
     verbosity_level: int = 0
+    additional_authorized_imports: list = field(default_factory=lambda: AUTHORIZED_IMPORTS)
+    planning_interval: int = 4
     # def __init__(self):
     def __post_init__(self):
         """Run post_init."""
@@ -86,8 +89,8 @@ class BasicAgent:
             tools=self.tools,
             model=self.model,
             verbosity_level=self.verbosity_level,
-            additional_authorized_imports=AUTHORIZED_IMPORTS,
-            planning_interval=4,
+            additional_authorized_imports=self.additional_authorized_imports,
+            planning_interval=self.planning_interval,
         )
 
     def get_answer(self, question: str):
@@ -122,53 +125,14 @@ def main():
 
     space_id = f"{username}/{repo_name}"
 
-    # model = get_model(cat="gemini")
-
-    _ = (
-        "gemini-2.5-flash-preview-04-17",
-        # "https://api-proxy.me/gemini/v1beta",
-        "https://generativelanguage.googleapis.com/v1beta",
-        os.getenv("GEMINI_API_KEY"),
-    )
-
-    _ = (
-        "grok-3-beta",
-        "https://api.x.ai/v1",
-        os.getenv("XAI_API_KEY"),
-    )
-
-    # model = litellm_model(*_)
-    model = openai_model(*_)
-
-    messages = [{'role': 'user', 'content': 'Say this is a test.'}]
-    print(model(messages))
-    # raise SystemExit("By intention")
-
-    # 1. Instantiate Agent ( modify this part to create your agent)
-    try:
-        # agent = BasicAgent()
-        agent = BasicAgent(
-            model=model,
-            tools=[
-                DuckDuckGoSearchTool(),
-                VisitWebpageTool(),
-                WikipediaSearchTool(),
-                FinalAnswerTool(),
-            ]
-        )
-        agent.agent.visualize()
-    except Exception as e:
-        print(f"Error instantiating agent: {e}")
-        return f"Error initializing agent: {e}", None
-
     # In the case of an app running as a hugging Face space, this link points toward your codebase ( usefull for others so please keep it public)
     agent_code = f"https://huggingface.co/spaces/{space_id}/tree/main"
     print(agent_code)
 
-    # 2. Fetch Questions
+    # 2. Fetch Questions: fetch before openai_model() which my set proxy
     print(f"Fetching questions from: {questions_url}")
     try:
-        response = requests.get(questions_url, timeout=30)
+        response = requests.get(questions_url, timeout=120)
         response.raise_for_status()
         questions_data = response.json()
         if not questions_data:
@@ -186,40 +150,101 @@ def main():
         print(f"An unexpected error occurred fetching questions: {e}")
         return f"An unexpected error occurred fetching questions: {e}", None
 
-    # 3. Run your Agent
-    results_log = []
-    answers_payload = []
+    # model = get_model(cat="gemini")
 
-    print(f"Running agent on {len(questions_data)} questions...")
+    _ = (
+        "gemini-2.5-flash-preview-04-17",
+        # "https://api-proxy.me/gemini/v1beta",
+        "https://generativelanguage.googleapis.com/v1beta",
+        os.getenv("GEMINI_API_KEY"),
+    )
 
-    # for item in questions_data:
-    # for item in questions_data[-1:]:
-    for item in questions_data[14:15]:
-        task_id = item.get("task_id")
-        question_text = item.get("question")
-        if not task_id or question_text is None:
-            print(f"Skipping item with missing task_id or question: {item}")
-            continue
-        try:
-            submitted_answer = agent(question_text)
-            answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
-            results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted_answer})
-        except Exception as e:
-            print(f"Error running agent on task {task_id}: {e}")
-            results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": f"AGENT ERROR: {e}"})
+    _ = (
+        "grok-3-beta",
+        "https://api.x.ai/v1",
+        os.getenv("XAI_API_KEY"),
+    )
+
+    # model = litellm_model(*_)
+    # model = get_model()
+
+    model = openai_model()  # defautl llama4 scout
+
+    # messages = [{'role': 'user', 'content': 'Say this is a test.'}]
+    # print(model(messages))
 
-    if not answers_payload:
-        print("Agent did not produce any answers to submit.")
-        return "Agent did not produce any answers to submit.", pd.DataFrame(results_log)
+    # raise SystemExit("By intention")
+
+    mcp_searxng_params = StdioServerParameters(
+        **{
+          "command": "npx",
+          "args": [
+            "-y",
+            "mcp-searxng"
+          ],
+          "env": {
+            "SEARXNG_URL": "https://searx.dattw.eu.org"
+          }
+        }
+    )
 
-    # 4. Prepare Submission
-    submission_data = {"username": username.strip(), "agent_code": agent_code, "answers": answers_payload}  # noqa
-    status_update = f"Agent finished. Submitting {len(answers_payload)} answers for user '{username}'..."
-    print(status_update)
-    print(answers_payload)
+    # with ToolCollection.from_mcp(mcp_searxng_params, trust_remote_code=True) as searxng_tool_collection, ToolCollection.from_mcp(mcp_markitdown_params, trust_remote_code=True) as markitdown_tools:
+    with ToolCollection.from_mcp(mcp_searxng_params, trust_remote_code=True) as searxng_tool_collection:
+        # 1. Instantiate Agent ( modify this part to create your agent)
+        try:
+            # agent = BasicAgent()
+            agent = BasicAgent(
+                model=model,
+                tools=[
+                    *searxng_tool_collection.tools,
+                    # DuckDuckGoSearchTool(),
+                    VisitWebpageTool(),
+                    WikipediaSearchTool(),
+                    FinalAnswerTool(),
+                ],
+                verbosity_level=1,
+            )
+            agent.agent.visualize()
+        except Exception as e:
+            print(f"Error instantiating agent: {e}")
+            return f"Error initializing agent: {e}", None
+
+
+        # 3. Run your Agent
+        results_log = []
+        answers_payload = []
+
+        print(f"Running agent on {len(questions_data)} questions...")
+
+        # for item in questions_data:
+        # for item in questions_data[-1:]:
+        # for item in questions_data[14:15]:
+        for item in questions_data[-6:]:
+            task_id = item.get("task_id")
+            question_text = item.get("question")
+            if not task_id or question_text is None:
+                print(f"Skipping item with missing task_id or question: {item}")
+                continue
+            try:
+                submitted_answer = agent(question_text)
+                answers_payload.append({"task_id": task_id, "submitted_answer": submitted_answer})
+                results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": submitted_answer})
+            except Exception as e:
+                print(f"Error running agent on task {task_id}: {e}")
+                results_log.append({"Task ID": task_id, "Question": question_text, "Submitted Answer": f"AGENT ERROR: {e}"})
+
+        if not answers_payload:
+            print("Agent did not produce any answers to submit.")
+            return "Agent did not produce any answers to submit.", pd.DataFrame(results_log)
+
+        # 4. Prepare Submission
+        submission_data = {"username": username.strip(), "agent_code": agent_code, "answers": answers_payload}  # noqa
+        status_update = f"Agent finished. Submitting {len(answers_payload)} answers for user '{username}'..."
+        print(status_update)
+        print(answers_payload)
 
-    agent.agent.visualize()
-    return None, None
+        agent.agent.visualize()
+        return None, None
 
 if __name__ == "__main__":
     main()

openai_model.py

@@ -26,7 +26,12 @@ def openai_model(
 
     # default llama4
     api_base = api_base or "https://api.llama.com/compat/v1"
+
     api_key = api_key or os.getenv("LLAMA_API_KEY")
+    if isinstance(api_key, str):
+        # LLAMA_API_KEY contains | and in win10 need to assign env var with ""
+        api_key = api_key.strip('"')
+    assert api_key, "LLAMA_API_KEY not set, set it and try again"
 
     # "Llama-4-Maverick-17B-128E-Instruct-FP8"
     # "Llama-4-Scout-17B-16E-Instruct-FP8"
@@ -43,10 +48,11 @@ def openai_model(
 def main():
     messages = [{'role': 'user', 'content': 'Say this is a test.'}]
     logger.debug(sys.argv)
+
     if not sys.argv[1:]:
+        logger.debug("default llama4 scout")
         model = openai_model()
         logger.debug(model(messages))
-        return
 
     if len(sys.argv[1:]) < 3:
         raise SystemExit("Provide at least three args (model_id, api_base, api_key)")