"""News Summarizer Agent
A Bindu agent that searches and summarizes latest news on any topic.
Runs locally via Ollama - no API key required!
Features:
- Real-time web search using DuckDuckGo
- Local LLM via Ollama
- Works for any topic: sports, tech, finance, politics etc..
Usage:
python news_agent.py
"""
import os
from dotenv import load_dotenv
from agno.agent import Agent
from agno.models.ollama import Ollama
from agno.tools.duckduckgo import DuckDuckGoTools
from bindu.penguin.bindufy import bindufy
load_dotenv()
# Agent Brain
agent = Agent(
instructions=(
"You are a news research assistant. "
"When given a topic, search for the latest news about it. "
"Return a structured summary with: "
"1. Top 3 headlines "
"2. Brief summary of each "
"3. Overall sentiment (positive/negative/neutral)"
),
model=Ollama(id="llama3.2"),
tools=[DuckDuckGoTools()],
)
# Bindu Config
config = {
"author": "SSJ.shabdsnehi08@gmail.com",
"name": "news_summarizer_agent",
"description": "Searches and summarizes latest news on any topic using local Ollama",
"deployment": {
"url": "http://localhost:3773",
"expose": True,
"cors_origins": ["http://localhost:5173"]
},
"skills": [],
}
# Handler
def handler(messages: list[dict[str, str]]):
"""Process incoming messages and return news summary.
Args:
messages: Conversation history as list of dicts
e.g. [{"role": "user", "content": "cricket news"}]
Returns:
Agent response with structured news summary
"""
latest_message = messages[-1]["content"]
result = agent.run(input=messages)
return result
# Launch
if __name__ == "__main__":
os.environ["AUTH_ENABLED"] = "false"
bindufy(config, handler)
