
Hallucinations Detector is a free and open-source tool that helps you verify the accuracy of your content instantly. Think of it as Grammarly, but for factual accuracy instead of grammar. It analyzes your content, identifies potential inaccuracies, and suggests corrections backed by reliable web sources.
Claim Extraction: When you input your content, the tool uses an LLM (Claude 3.5 Sonnet) to break down your text into individual claims.
Source Verification: Each claim is checked using Exa’s search tool to find reliable sources online that either support or refute it.
Accuracy Analysis: The claims and their corresponding sources are analyzed by our LLM to determine their accuracy.
Results Display: Finally, we show the results in a simple, clear way, pointing out any mistakes and offering suggestions to fix them.
git clone https://github.com/exa-labs/exa-hallucination-detector.git
cd exa-hallucination-detector
````
2. Install dependencies
npm install
yarn install
3. Set up environment variables Create a `.env.local` file in the root directory and add your API keys:
EXA_API_KEY=your_exa_api_key ANTHROPIC_API_KEY=your_anthropic_api_key
4. Run the development server
npm run dev
yarn dev ```
Get your Exa API key from Exa Dashboard
Get your Anthropic API key from Anthropic Documentation
This project is powered by Exa.ai, a cutting-edge search engine designed specifically for AI applications. Exa provides:
Advanced semantic and keyword-based search capabilities
Instant retrieval of clean web content
Customizable search parameters
Similarity search using URLs or text
Superior search capabilities compared to traditional search APIs
Built with ❤️ by team Exa
$ claude mcp add exa-hallucination-detector \
-- python -m otcore.mcp_server <graph>