Publish a widget, chatbot, feed, or MCP endpoint

Publishing makes selected library material available to people outside your workspace. Review the publishing settings and confirm what will become public before enabling it. Choose whether document text, source metadata, and other fields may be exposed.

A search widget lets visitors find published documents. A chatbot lets them ask questions. Copy the supplied JavaScript into your website and configure its allowed domains. Your website's security policy must allow the ARXIVIX script and network requests. Test it on the actual destination website.

A structured feed provides published records as JSON. An MCP endpoint lets a compatible AI assistant retrieve published material at question time. Copy the endpoint from your publishing settings into a client that supports it. This is retrieval, not permission to train a model on your content.

Questions asked through a public chatbot or an MCP ask tool share the publishing workspace's monthly question allowance. Reaching that allowance pauses new answers without an automatic overage charge. Ordinary browsing and metadata retrieval do not create a separate question allowance.

Metadata-only publishing does not provide document text or excerpt answers. Unpublish a library or disable an embed when you no longer want it accessible. Copies already downloaded by others may remain outside ARXIVIX.