xrad provides a structured claim store that enforces verbatim evidence for every assertion made by LLM agents. Designed for extraction pipelines and knowledge-graph construction, it ensures agents cannot hallucinate or assert without quoting their source. It includes MCP support and is licensed under Apache-2.0, making it suitable for building trustworthy AI systems that require strong provenance and citation guarantees.
In the LLM eval & observability space, xrad takes a focused approach. LLM agents making unsubstantiated claims without proper source citation in extraction and reasoning pipelines. xrad is an open-source project aimed at AI developers and LLM pipeline engineers. xrad is open source under the Apache-2.0 license. It runs on the command line and API, and it can be self-hosted.
xrad first shipped in 2026. Among its 5 catalogued features are Verbatim Evidence Enforcement, Claim Store, and Hallucination Prevention. It exposes integrations via an MCP server.
Summary written by a language model from the project’s public pages.
What PulseGate has recorded for this listing
Same category — not a similarity match