From office-docx-skills
Use when a `.docx` or Word task requires paragraph-by-paragraph translation QA, format review, bilingual document audit, formula preservation checks, 逐段核对, 翻译审校, or validation before deleting Chinese text.
How this skill is triggered — by the user, by Claude, or both
Slash command
/office-docx-skills:docx-format-auditThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Audit bilingual or English-only Word translation outputs before finalization. This skill checks structure and formatting; it does not perform translation by itself.
Audit bilingual or English-only Word translation outputs before finalization. This skill checks structure and formatting; it does not perform translation by itself.
Use this skill when:
Do not use this skill for non-DOCX code, finance, or general audit tasks.
| Need | Required action |
|---|---|
| Bilingual translation QA | Match btx_<id>_src to btx_<id>_en; missing pair markers are blocking |
| Table-cell review | Require source and English pair in the same w:tc |
| Formula preservation | Compare per-pair OMML, drawing, VML, and OLE signatures |
| English typography check | Confirm inserted English resolves to Times New Roman |
| English-only cleanup | Run audit first, remove source paragraphs, then audit with --mode english-only |
The RED baseline capture must show at least one omitted guardrail before this skill is created. This skill closes the observed baseline gaps for DOCX audit work: deleting Chinese text without pair-ID checks, accepting missing markers, auditing only body paragraphs while missing tables, counting formulas at document level instead of per pair, or accepting English-only output that still contains Chinese text. If a future baseline covers one of these without the skill, remove that claim from this section instead of keeping untested guidance.
Fail the audit when:
Times New Roman through direct formatting or style inheritance;python-docx.Warn on:
the the;Use stable pair IDs. Missing pair markers are blocking because adjacency is not safe enough for automated audit or deletion. Formula preservation should compare per-pair signatures, including object type, relationship target, dimensions, media hash, file name, and paragraph/table-cell context.
From the package root, this verifies that the audit helper is importable:
python3 - <<'PY'
import importlib.util
import sys
from pathlib import Path
audit_path = Path("skills/docx-format-audit/audit_docx_translation.py")
spec = importlib.util.spec_from_file_location("audit_docx_translation", audit_path)
assert spec is not None and spec.loader is not None
module = importlib.util.module_from_spec(spec)
sys.modules[spec.name] = module
spec.loader.exec_module(module)
assert callable(module.main)
print("PASS: audit helper import")
PY
docx-bilingual-translation.word-formula-writing for formula-heavy documents.word-default-formatting when typography and layout must be validated.| Mistake | Fix |
|---|---|
| Auditing only marked pairs | Also block unmarked Chinese body and table-cell paragraphs |
| Treating adjacency as proof | Treat missing pair IDs as blocking; add bookmarks before audit cleanup |
| Counting document-level formulas | Compare formula signatures for each source/English pair |
| Deleting Chinese before audit | Run bilingual audit first and refuse deletion on blocking findings |
| Ignoring English-only output | Re-run audit in English-only mode after removal |
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub wintersdragon-c/office-docx-skills --plugin office-docx-skills