Querying Classical Chinese Medicine by Symptom, Formula, and Screenshot

What it does

nihaisha is a Claude Code / Codex skill that turns Ni Haisha’s extensive TCM lecture corpus into a structured, queryable reference. It maps colloquial symptoms like “cold with chills” to course-specific diagnostic frameworks, then surfaces relevant formulas, acupoints, and lecture screenshots from modules spanning Shang Han Lun, Jin Gui Yao Lue, acupuncture, Huang Di Nei Jing, and more. Answers are backed by page-level PDF evidence cards rather than left to the model’s imagination.

The interesting bit

Most agent skills are text prompts with a few JSON files; this one ships with nearly 3,000 compressed WebP screenshots and 11 extractable PDFs stored directly in the repo. You can ask for “screenshot evidence of Xiao Chai Hu Tang” and receive actual image paths, or cite a source as pdf-evidence:<doc_id>#p<page>—making retrieval visual, specific, and auditable.

Key highlights

• Covers 13 course modules including Shang Han Lun, Jin Gui Yao Lue, Tian Ji, and the Shen Nong Ben Cao Jing
• Indexes 2,986 lecture screenshots and 3,003 PDF page-level evidence cards for traceable retrieval
• Translates plain-language symptoms into TCM diagnostic “watershed” questions before fetching course content
• Explicitly walled off from clinical use: no diagnosis, prescriptions, or dosage guidance
• Distilled using the author’s lineage-skill methodology for converting dense video courses into agent-native formats

Caveats

• The README warns the project is still iterating and recommends checking back for updates before relying on it
• Two modules (the Liang Dong interview and Stanford lecture) are text-only and lack screenshot evidence
• It is strictly a study aid; the safety notice repeatedly disclaims any therapeutic or medical-decision use

Verdict

Grab it if you are studying Ni Haisha’s curriculum and want an agent that can cite specific lectures and whiteboard screenshots. Skip it if you need a medical diagnostic tool or a polished, finished product.