medical-literature-translationlisted
Install: claude install-skill qitong/medical-literature-translation
# Medical Literature Translation
Produce translations a domain peer-reviewer would accept: scientifically exact first,
idiomatic and elegant second. Default output is **translation-only** in the target language.
## When this runs
Translate medical/biomedical/clinical text between Chinese and English. If the direction
is not stated, infer it from the source language and confirm in one line before a long job.
## Core workflow
### 1. Intake
- Identify **direction** (ZH→EN or EN→ZH) and **source text** (pasted, or a file — for
`.pdf`/`.docx`, extract text first via the `pdf`/`docx` skills).
- Ask for or accept an optional **per-job glossary** (any format, or the CSV template in
`assets/glossary-template.csv`). Absence is fine.
### 2. Load terminology
- Read `references/glossary.md` (persistent term library — always applied).
- Merge the per-job glossary on top; **per-job overrides persistent** on any conflict.
- Match terms case-insensitively as whole words/phrases; **longest match wins**.
### 3. Mark protected (do-not-translate) tokens
- Run `scripts/find_protected_terms.py <sourcefile>` (or `--json`) to list every token that
must stay verbatim: references, DOIs, URLs, PMIDs, trial IDs, citation markers, units,
statistics, gene/biomarker symbols. Keep this list as the verbatim-preservation checklist.
- Rules for what stays verbatim (reference titles, links, symbols, etc.) are in
`references/translation-conventions.md` §2. **Read that file before translating.**