File size: 3,037 Bytes
a2913a9 3b69291 a2913a9 3b69291 c3c6e00 131111c c3c6e00 3b69291 99137ea 5997785 3b69291 131111c 3b69291 99137ea 3b69291 99137ea 3b69291 99137ea c3c6e00 3b69291 5997785 99137ea 3b69291 c3c6e00 3b69291 99137ea 3b69291 866c379 3b69291 5997785 a2913a9 3b69291 a2913a9 3b69291 5997785 3b69291 c3c6e00 99137ea |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 |
"""
Gradio front-end + smolagents CodeAgent
---------------------------------------
• If you set an OPENAI_API_KEY the agent will call OpenAI (GPT-4o by default).
• Otherwise it falls back to a free Hugging Face chat-completion model
(defaults to microsoft/Phi-3-mini-4k-instruct which the public Inference
API exposes for the Chat Completion task).
• You can override the fallback by defining HF_MODEL_ID and, if needed,
HF_API_TOKEN in the Space → Settings → Secrets.
"""
import os, pathlib, gradio as gr
from mcp import StdioServerParameters
from smolagents import MCPClient, CodeAgent, InferenceClientModel
# Path to your MCP tool server
SERVER_PATH = pathlib.Path(__file__).with_name("mcp_server.py")
# Decide which base model to use
OPENAI_KEY = os.getenv("OPENAI_API_KEY")
HF_MODEL_ID = os.getenv("HF_MODEL_ID", "microsoft/Phi-3-mini-4k-instruct")
if OPENAI_KEY: # --- OpenAI branch -------------
from smolagents.models import OpenAIChat # <- works in all versions
BASE_MODEL = OpenAIChat( # gpt-4o by default
model=os.getenv("OPENAI_MODEL", "gpt-4o-preview"),
temperature=0.3,
)
else: # --- Hugging Face branch ----
BASE_MODEL = InferenceClientModel( # uses HF Inference API
model_id=HF_MODEL_ID,
hf_api_token=os.getenv("HF_API_TOKEN"), # optional for gated repos
timeout=90,
)
# ----------------- callback ---------------------------------------------------
def respond(message: str, history: list):
"""Send user prompt → CodeAgent → SQL tools → natural-language answer."""
params = StdioServerParameters(command="python", args=[str(SERVER_PATH)])
with MCPClient(params) as tools:
agent = CodeAgent(tools=tools, model=BASE_MODEL)
answer = agent.run(message)
history += [
{"role": "user", "content": message},
{"role": "assistant", "content": answer},
]
return history, history
# ----------------- UI ---------------------------------------------------------
with gr.Blocks(title="Enterprise SQL Agent") as demo:
state = gr.State([])
gr.Markdown("## Enterprise SQL Agent — ask questions about your data 🏢➡️📊")
chat = gr.Chatbot(type="messages", label="Chat")
box = gr.Textbox(
placeholder="e.g. Who are my inactive Northeast customers?",
show_label=False,
)
box.submit(respond, [box, state], [chat, state])
with gr.Accordion("Example prompts"):
gr.Markdown(
"* Who are my **Northeast** customers with no orders in 6 months?\n"
"* List customers sorted by **LastOrderDate**.\n"
"* Draft re-engagement emails for inactive accounts."
)
footer = (
f"_Powered by MCP + smolagents + Gradio • Model: "
f"{'OpenAI' if OPENAI_KEY else HF_MODEL_ID}_"
)
gr.Markdown(footer)
if __name__ == "__main__":
demo.launch()
|