ittia-research/check

Extracts claims from text, searches the web for evidence via search engines, and uses LLM-powered analysis to assign verdicts (true/false/tie/irrelevant) with citations from top results. The architecture combines retrieval through customizable search backends, content extraction, multi-source ranking via embedding and reranking models, and LLM verdict generation—supporting self-hosting with OpenAI-compatible APIs, Ollama, or vLLM. Outputs structured analyses in Markdown or JSON, treating unverified claims as doubtful by default while distinguishing factual statements from opinions and questions.

No commits in the last 6 months.