Initial QQwrite skill

SKILL.md

@@ -0,0 +1,75 @@
+---
+name: qqwrite
+description: Adjust manuscript Word documents to match journal or group writing templates. Use when the user asks QQwrite to format, restructure, clean, or align a .docx manuscript according to a Word template from the local template library, especially for scientific manuscripts, AFM-style templates, section order, title page metadata, abstract/keywords placement, headings, captions, references, SI pointers, and submission-ready Word formatting.
+---
+
+# QQwrite
+
+QQwrite is the Word-template execution skill for manuscript writing workflows. Its current scope is intentionally narrow: use a template from the local template library to adjust a user's manuscript Word file while preserving scientific meaning.
+
+## Core Rule
+
+Do not overwrite the source manuscript. Always create a revised copy and a short Markdown report.
+
+Default outputs:
+
+- `*_qqwrite_adjusted.docx`
+- `*_qqwrite_report.md`
+
+## Template Library
+
+Use bundled templates first:
+
+- `assets/templates/afm/article-template.docx` for Advanced Functional Materials-style manuscript formatting.
+
+If the user provides another template path, use the user-provided template instead.
+
+For template handling details, read `references/template-adjustment.md`.
+
+## Workflow
+
+1. Identify inputs:
+   - source manuscript `.docx`
+   - template `.docx`
+   - target journal or template name if provided
+   - output directory
+
+2. Preflight both documents:
+   - inspect section order and heading hierarchy
+   - compare title page fields, abstract, keywords, main text, methods, references, captions, acknowledgements, and supporting information pointers
+   - inspect paragraph styles, table count, figure-caption patterns, references section, and obvious empty or duplicated sections
+   - run `scripts/docx_template_audit.py` when a quick structural report is useful
+
+3. Adjust only formatting and template structure unless the user explicitly asks for rewriting:
+   - move or relabel sections to match the template
+   - normalize heading levels
+   - normalize title, author, affiliation, abstract, keywords, captions, references, and acknowledgements placement
+   - preserve scientific claims, numbers, units, equations, citations, and figure/table labels
+   - flag unclear template conflicts in the report instead of guessing silently
+
+4. Produce the revised Word file:
+   - copy the original manuscript first
+   - apply template-compatible styles and section order
+   - keep original content recoverable
+   - avoid converting citations to plain text unless unavoidable and reported
+
+5. Produce the report:
+   - list template used
+   - list changed sections and formatting changes
+   - list unresolved issues requiring author decision
+   - list any content that appears missing relative to the template
+
+## Boundaries
+
+- QQwrite does not decide manuscript novelty, literature strength, or reviewer risk. Route those tasks to QQsci.
+- QQwrite does not generate Zotero/DeepSeek literature notes. Route those tasks to QQnote.
+- QQwrite may lightly fix obvious Word-format inconsistencies, but substantial scientific rewriting belongs in QQsci or a future writing module expansion.
+- For citation suggestions inside Word, follow QQsci's DOI-only comment convention if the task explicitly involves citation comments.
+
+## Current Capability
+
+Only one capability is active now:
+
+`template-adjust-docx`: adjust a manuscript `.docx` according to a `.docx` template from the local library.
+
+Future capabilities can be added later, but keep this skill focused on Word-template execution.

agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: "QQwrite"
+  short_description: "Adjust manuscript Word documents to match writing templates."
+  default_prompt: "Help me adjust this manuscript Word document using a template from the template library. Preserve scientific meaning, compare structure and formatting with the template, and produce a revised .docx plus a short Markdown report."

references/template-adjustment.md

@@ -0,0 +1,62 @@
+# Word Template Adjustment
+
+## Goal
+
+Make the user's manuscript follow the selected Word template without changing the scientific meaning.
+
+## What To Compare
+
+Compare these elements before editing:
+
+- title page fields: title, authors, affiliations, corresponding author, abstract, keywords
+- section order: introduction, results/discussion, conclusion, experimental/methods, acknowledgements, conflict of interest, data availability, references
+- heading hierarchy and visible heading text
+- figure and table caption style and numbering
+- references section placement and numbering style
+- SI references, graphical abstract/TOC notes, highlights, and cover letter cross-references when present
+- page setup, margins, columns, headers/footers, and line spacing when detectable
+
+## Editing Rules
+
+- Work on a copied `.docx`; never overwrite the source.
+- Prefer Word-native styles when available.
+- Preserve all numbers, units, chemical formulas, gene/protein names, sample names, equations, citation markers, figure labels, and table labels.
+- Do not silently delete content that has no obvious template location. Move it to the closest suitable section or flag it in the report.
+- If the template contains placeholder text, replace only with matching manuscript content; remove unused placeholders in the revised copy.
+- Keep references and citations intact. If field codes are lost by tooling, report that explicitly.
+
+## Report Format
+
+Use this Markdown shape:
+
+```markdown
+# QQwrite Template Adjustment Report
+
+## Inputs
+- Manuscript:
+- Template:
+- Output:
+
+## Changes Made
+- ...
+
+## Missing Or Unclear Items
+- ...
+
+## Checks
+- Section order:
+- Heading hierarchy:
+- Captions:
+- References:
+- Citation fields:
+```
+
+## When To Stop And Ask
+
+Ask the user before continuing if:
+
+- the manuscript and template are for clearly different article types
+- the manuscript has no recognizable title/abstract/main sections
+- a template-required section is missing and cannot be inferred
+- editing would require scientific rewriting rather than formatting/structure adjustment
+- citation fields, equations, or embedded objects are at risk of being flattened

scripts/docx_template_audit.py

@@ -0,0 +1,127 @@
+#!/usr/bin/env python3
+"""Create a lightweight structural audit for a manuscript/template DOCX pair.
+
+This script intentionally uses only the Python standard library. It inspects the
+OOXML package directly and reports headings, paragraph-style usage, section-like
+paragraphs, table count, and image count. It does not modify documents.
+"""
+
+from __future__ import annotations
+
+import argparse
+import collections
+import json
+import re
+import sys
+import zipfile
+from pathlib import Path
+from xml.etree import ElementTree as ET
+
+
+NS = {
+    "w": "http://schemas.openxmlformats.org/wordprocessingml/2006/main",
+    "a": "http://schemas.openxmlformats.org/drawingml/2006/main",
+}
+
+
+SECTION_PATTERNS = [
+    "abstract",
+    "keywords",
+    "introduction",
+    "results",
+    "discussion",
+    "conclusion",
+    "experimental",
+    "methods",
+    "acknowledgements",
+    "acknowledgments",
+    "conflict of interest",
+    "data availability",
+    "references",
+]
+
+
+def read_xml(zf: zipfile.ZipFile, name: str) -> ET.Element | None:
+    try:
+        return ET.fromstring(zf.read(name))
+    except KeyError:
+        return None
+
+
+def text_from_paragraph(p: ET.Element) -> str:
+    parts = []
+    for t in p.findall(".//w:t", NS):
+        if t.text:
+            parts.append(t.text)
+    return "".join(parts).strip()
+
+
+def style_from_paragraph(p: ET.Element) -> str:
+    style = p.find("./w:pPr/w:pStyle", NS)
+    if style is None:
+        return "(none)"
+    return style.attrib.get(f"{{{NS['w']}}}val", "(unknown)")
+
+
+def inspect_docx(path: Path) -> dict:
+    with zipfile.ZipFile(path) as zf:
+        document = read_xml(zf, "word/document.xml")
+        if document is None:
+            raise ValueError(f"{path} does not contain word/document.xml")
+
+        paragraphs = document.findall(".//w:p", NS)
+        tables = document.findall(".//w:tbl", NS)
+        drawings = document.findall(".//w:drawing", NS)
+
+        style_counts: collections.Counter[str] = collections.Counter()
+        headings = []
+        section_hits = []
+        nonempty_count = 0
+
+        for index, p in enumerate(paragraphs, start=1):
+            text = text_from_paragraph(p)
+            style = style_from_paragraph(p)
+            style_counts[style] += 1
+            if not text:
+                continue
+            nonempty_count += 1
+            normalized = re.sub(r"\s+", " ", text).strip().lower()
+            if style.lower().startswith("heading") or style.lower().startswith("title"):
+                headings.append({"index": index, "style": style, "text": text[:160]})
+            if any(pattern == normalized or normalized.startswith(pattern + ":") for pattern in SECTION_PATTERNS):
+                section_hits.append({"index": index, "style": style, "text": text[:160]})
+
+        return {
+            "path": str(path),
+            "paragraphs": len(paragraphs),
+            "nonempty_paragraphs": nonempty_count,
+            "tables": len(tables),
+            "drawings": len(drawings),
+            "top_styles": style_counts.most_common(20),
+            "headings": headings[:80],
+            "section_hits": section_hits[:80],
+        }
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser(description="Audit a manuscript DOCX against a template DOCX.")
+    parser.add_argument("--manuscript", required=True, type=Path)
+    parser.add_argument("--template", required=True, type=Path)
+    parser.add_argument("--out", type=Path, help="Optional JSON output path.")
+    args = parser.parse_args()
+
+    report = {
+        "manuscript": inspect_docx(args.manuscript),
+        "template": inspect_docx(args.template),
+    }
+
+    payload = json.dumps(report, ensure_ascii=False, indent=2)
+    if args.out:
+        args.out.write_text(payload, encoding="utf-8")
+    else:
+        print(payload)
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())