LingJing

LingJing runs a local virtual classroom where the assistant acts as a questioning AI student that probes a topic you explain and returns blindspots and a…

What it does

LingJing is a self-hosted virtual classroom: a human explains something they know, and one AI student (a mirror) asks follow-up questions from seven probe categories. A deterministic layer decides what to ask and which weak point to target, while an LLM only phrases the questions. After class it returns a side-by-side list of what was asked and what you answered, plus blindspots and a summary. It explicitly does not answer questions, grade, or judge.

When to use it

Useful for self-testing comprehension of material you already know, extracting tacit experience into queryable form, team handover, teaching practice, expert interviews, or research comparison. It is meant for local or intranet self-hosting, not as a public multi-tenant service.

Requirements

Local process, Node.js 22 or newer, no third-party dependencies; runs via node server.js and is opened in a browser at localhost. No account or authentication. Optional LLM phrasing requires an API key via environment variables such as LINGJING_OR_KEY, LINGJING_ZHIPU_KEY, LINGJING_SILICONFLOW_KEY, or LINGJING_DEEPSEEK_KEY, selected with LINGJING_LLM_PROVIDER. Without a key it runs on deterministic fallback text with no outbound calls.