feat: adding api key usage instead of using the default api key.

src/insurance_assistants/agents.py

@@ -1,8 +1,7 @@
+# from dotenv import find_dotenv, load_dotenv
 from pathlib import Path
 
-# from dotenv import find_dotenv, load_dotenv
-import os
-from huggingface_hub import InferenceClient, login
+from huggingface_hub import InferenceClient  # , login
 from smolagents import (
     CodeAgent,
     DuckDuckGoSearchTool,
@@ -24,6 +23,7 @@ rag_app = RAG()
 rag_app.vectordb_id = "policy_wordings"
 # login(os.getenv("HF_TOKEN"))
 
+
 class InsuranceInfoRetriever(Tool):
     name = "InsuranceInfoRetriever"
     description = "Retrieves information from insurance documents."
@@ -35,7 +35,7 @@ class InsuranceInfoRetriever(Tool):
     def forward(self, query: str) -> str:
         client = InferenceClient(
             provider="hyperbolic",
-            bill_to="VitalNest",
+            # bill_to="VitalNest",
             # token=os.getenv("HF_TOKEN")
         )
         results = rag_app.search_documents(query)
@@ -73,9 +73,11 @@ class InsuranceInfoRetriever(Tool):
 
 insurance_agent = CodeAgent(
     tools=[InsuranceInfoRetriever(), FinalAnswerTool()],
-    model=InferenceClientModel(bill_to="VitalNest", temperature=0.1,
-                              # token=os.getenv("HF_TOKEN")
-                              ),
+    model=InferenceClientModel(
+        # bill_to="VitalNest",
+        temperature=0.1,
+        # token=os.getenv("HF_TOKEN")
+    ),
     additional_authorized_imports=["os", "requests", "bs4", "pil", "base64", "io"],
     max_steps=1,
     verbosity_level=-1,
@@ -85,7 +87,9 @@ insurance_agent = CodeAgent(
 )
 websearch_agent = ToolCallingAgent(
     model=InferenceClientModel(
-        model_id="Qwen/Qwen3-30B-A3B", bill_to="VitalNest", temperature=0.1,
+        model_id="Qwen/Qwen3-30B-A3B",
+        # bill_to="VitalNest",
+        temperature=0.1,
         # token=os.getenv("HF_TOKEN")
     ),
     tools=[
@@ -102,7 +106,8 @@ websearch_agent = ToolCallingAgent(
 
 wikipedia_agent = ToolCallingAgent(
     model=InferenceClientModel(
-        model_id="Qwen/Qwen3-30B-A3B", bill_to="VitalNest", temperature=0.1,
+        model_id="Qwen/Qwen3-30B-A3B",
+        # bill_to="VitalNest", temperature=0.1,
         # token=os.getenv("HF_TOKEN")
     ),
     tools=[
@@ -120,7 +125,7 @@ manager_agent = CodeAgent(
     additional_authorized_imports=["os"],
     model=InferenceClientModel(
         model_id="Qwen/Qwen3-235B-A22B",
-        bill_to="VitalNest",
+        # bill_to="VitalNest",
         temperature=0.1,
         # token=os.getenv("HF_TOKEN")
     ),
@@ -134,4 +139,3 @@ manager_agent = CodeAgent(
     "health insurance documents using the `insurance_agent` team member, search wikipedia and the web for general information.",
 )
 manager_agent.system_prompt = manager_agent.system_prompt + PROMPT_PREFIX
-

src/insurance_assistants/complex_rag.py

@@ -16,6 +16,7 @@ from colpali_engine.models import (
     ColQwen2_5_Processor,
 )
 from colpali_engine.utils.torch_utils import ListDataset, get_torch_device
+
 # from dotenv import find_dotenv, load_dotenv
 # from openai import OpenAI
 from pdf2image import convert_from_path

src/insurance_assistants/ui.py

@@ -5,16 +5,19 @@ import re
 import shutil
 
 import gradio as gr
-from dotenv import load_dotenv
 from gradio_pdf import PDF
+from huggingface_hub import login
 from smolagents.gradio_ui import _process_action_step, _process_final_answer_step
 from smolagents.memory import ActionStep, FinalAnswerStep, MemoryStep, PlanningStep
 from smolagents.models import ChatMessageStreamDelta
-from huggingface_hub import login
 
 # from smolagents import CodeAgent, InferenceClientModel
 from src.insurance_assistants.agents import manager_agent
-from src.insurance_assistants.consts import PRIMARY_HEADING, PROJECT_ROOT_DIR, PROMPT_PREFIX
+from src.insurance_assistants.consts import (
+    PRIMARY_HEADING,
+    PROJECT_ROOT_DIR,
+    PROMPT_PREFIX,
+)
 
 # load_dotenv(override=True)
 
@@ -90,11 +93,15 @@ class UI:
     def interact_with_agent(self, prompt, messages, session_state, api_key):
         # Get or create session-specific agent
         if not api_key or not api_key.startswith("hf"):
-            raise ValueError("Incorrect API key")
+            raise ValueError("Incorrect HuggingFace Inference API Key")
+        # Login to Hugging Face with the provided API key
+        login(token=api_key)
         if "agent" not in session_state:
             # session_state["agent"] = CodeAgent(tools=[], model=InfenceClientModel())
             session_state["agent"] = manager_agent
-        session_state["agent"].system_prompt = session_state["agent"].system_prompt + PROMPT_PREFIX
+        session_state["agent"].system_prompt = (
+            session_state["agent"].system_prompt + PROMPT_PREFIX
+        )
 
         # Adding monitoring
         try:
@@ -223,9 +230,12 @@ class UI:
 """
                         )
                         with gr.Group():
-                            api_key = gr.Textbox(placeholder="Enter your HuggingFace Inference API KEY HERE", 
-                                                 label="API key",show_label=True, type='password')
-                            login(token=api_key)
+                            api_key = gr.Textbox(
+                                placeholder="Enter your HuggingFace Inference API KEY HERE",
+                                label="🤗 Inference API Key",
+                                show_label=True,
+                                type="password",
+                            )
                             gr.Markdown(
                                 value="**Your question, please...**", container=True
                             )
@@ -342,6 +352,8 @@ class UI:
         demo.queue(max_size=4).launch(debug=False, **kwargs)
 
 
-if __name__=="__main__":
-    UI().launch(share=True,
-allowed_paths=[(PROJECT_ROOT_DIR /"data/policy_wordings").as_posix()])
+# if __name__ == "__main__":
+#     UI().launch(
+#         share=True,
+#         allowed_paths=[(PROJECT_ROOT_DIR / "data/policy_wordings").as_posix()],
+#     )