feat: add cross-validated-search example for hallucination-free responses

[wd041216-bit]

Overview

This PR adds a new example demonstrating how to use cross-validated-search with Ollama to prevent hallucinations by verifying facts against multiple sources.

Features

• ✅ Confidence Scoring: Every fact gets a confidence level (Verified/Likely True/Uncertain/Likely False)
• ✅ Cross-Validation: Facts verified against multiple independent sources
• ✅ No API Key Required: Completely free, no API keys needed
• ✅ Seamless Integration: Works with Ollama's chat API out of the box

Example Output

🥚 cross-validated-search + Ollama
Prevent hallucinations with confidence-scored search results

Searching: What is the latest version of Python?

┌──────────────────────────────────────────────────────────────┐
│ Results for: What is the latest version of Python?           │
├─────────┬────────────────────────────────────────┬───────────┤
│ Engine  │ Title                                   │ Confidence│
├─────────┼────────────────────────────────────────┼───────────┤
│ DuckDuckGo│ Python 3.13.0 Released              │ verified  │
│ Bing     │ Download Python                       │ verified  │
│ Google   │ Python.org                            │ verified  │
└─────────┴────────────────────────────────────────┴───────────┘

LLM Response:
Based on verified sources, Python 3.13.0 is the latest stable release...

How It Works

1. User asks a factual question
2. cross-validated-search queries multiple search engines
3. Results are cross-validated and assigned confidence scores
4. Verified context is passed to Ollama's LLM
5. LLM responds with accurate, cited information

Installation

pip install cross-validated-search ollama rich

Related

• examples/web-search.py - Basic web search with Ollama
• examples/web-search-mcp.py - MCP server for web search

This example complements the existing web search examples by adding hallucination prevention capabilities.