agentsclimarketplace

Create cn official pdf

Skill mmhsdxwdxx/skills/skills/official/create-cn-official-pdf

Create, typeset, inspect, or correct Chinese Party and government official-document PDFs under GB/T 9704—2012 and the Regulations on Handling Official Documents. Use for 通知、请示、报告、函、批复、纪要、决定、决议、命令(令)、公报、公告、通告、意见、议案、通报, 红头文件, 公文排版, 公文格式核验, or requests involving Chinese official-document fonts, font sizes, margins, line grids, indents, spacing, page numbers, seals, attachments, copy recipients, printing notes, PDF generation, or PDF preflight.From its SKILL.md

Install
npx -y skills add mmhsdxwdxx/skills --skill create-cn-official-pdf

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • 24 days oldThe repository was created 24 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

5.5 KB, ~1.1k tokens by cl100k_base, as published. Nobody here has run it

Create Chinese Official-Document PDFs

Produce a print-ready PDF that meets the national format baseline—not text that merely resembles a government document. Treat GB/T 9704—2012 as the national baseline, and treat any superior rule, issuing-authority template, or supplied red-head/seal asset as an explicit override recorded before generation. For choosing the document type, drafting content, or reviewing wording, use $draft-cn-official-document; this skill typesets and verifies the format.

Mandatory source loading

  1. Read references/format-standard.md in full before typesetting or auditing.
  2. When the user supplies an agency template or local rule, inspect it and record every deviation from the national baseline before generation.

Workflow

  1. From the request or source material, determine the document type, direction of communication, issuing authority, recipients, document number, signers, attachments, date, seal status, copy recipients, and printing authority/date.

  2. Derive the layout from the direction and type: general by default; upward for a 请示 or 报告 addressed to a superior; letter (信函格式) when the competent office authorizes it; order (命令/令); minutes (纪要). Never invent an issuing authority, document number, signer, or seal.

  3. Normalize the content into the JSON structure shown in assets/example.json. Preserve official names, dates, citations, punctuation, and attachment titles exactly.

  4. Generate the PDF with the bundled runtime:

    python scripts/official_pdf.py input.json output.pdf --report output.layout.json --render-dir rendered
    
  5. Read the layout report. Treat a missing 小标宋, a substituted font, missing required metadata, overflow, and an omitted seal as surfaced warnings—never as silent fallbacks.

  6. Inspect every rendered PNG. Verify red elements, title wrapping, first-page body presence, 22-line grid behavior, two-character indents, signatures, attachments, the edition record, odd/even page numbers, and the absence of clipping or stray blank pages.

  7. After the final edit, run the verifier:

    python scripts/verify_official_pdf.py output.pdf --report output.layout.json
    
  8. Deliver the PDF with a concise compliance note listing any agency overrides or unresolved warnings.

Non-negotiable rules

  • Page: A4 with a 156 mm × 225 mm type area, a 37 mm ± 1 mm top white margin, and a 28 mm ± 1 mm binding-side margin, unless a special format explicitly overrides placement.
  • Fonts by element: title in 2号小标宋; ordinary body in 3号仿宋; level-1 heading 一、 in 3号黑体, level-2 (一) in 3号楷体, level-3 1. and level-4 (1) in 3号仿宋; edition record in 4号仿宋; page numbers in 4号宋体.
  • Body spacing: no paragraph-before or paragraph-after spacing. Express vertical gaps in whole “lines”; keep the body on the 22-line grid with each ordinary line holding 28 full-width cells.
  • Indentation: every natural body paragraph starts two full-width cells in; wrapped lines return flush left. Titles and recipient lines are not indented.
  • Seals: do not synthesize, trace, or imitate an official seal. Place only a user-supplied authorized seal image; otherwise reserve the space and flag the omission.
  • Font substitutes: never claim a substitute is compliant. If 小标宋 is unavailable, use the closest readable Song-style face only for a draft and report it.
  • Line spacing: there is no universal Word “fixed line spacing” value. The national standard defines a 22-line page and a standard line unit; values such as 28.8, 29, or 30 pt are implementation choices that may need calibration to the installed font.
  • Digits and Latin letters: keep them in the surrounding prescribed font unless the user or an agency rule explicitly requires Times New Roman. Times New Roman is a common local convention, not a blanket GB/T 9704—2012 mandate.
  • Output: keep PDF text searchable and embed fonts when licensing and the renderer permit it.

Script behavior

The generator requires reportlab and the verifier requires pypdf; page rendering prefers pdftoppm and falls back to PyMuPDF. Pin them with requirements.txt.

official_pdf.py controls millimetre coordinates directly and does not rely on Word pagination. It supports the five layout modes—general, upward, letter, order, minutes—plus body paragraphs, four heading levels, attachments, signature/date blocks, optional authorized seal images, edition records, and odd/even page numbering. Pass --strict-fonts for final production when the exact required fonts are installed.

The generator is a standards-oriented baseline. For complex tables, landscape inserts, multi-authority seals, minority-language documents, or an agency-specific red-head asset, adapt the script or use an approved template, then visually verify every page.

What ships with it: 7 files

45.9 KB alongside SKILL.md, 3 of them executable

agents/

assets/

references/

Gives 0 of the 12 instructions most pdf office docs skills give in ~1.1k tokens

Counted across 636 of the 690 authors here whose files we hold, read 2026-08-07

  • Extract text or tables using pdfplumber or pdftotextin 89 of 636, across 23 files
  • Create new PDFs using reportlabin 83 of 636, across 16 files
  • Read forms.md before filling out PDF formsin 80 of 636, across 13 files
  • OCR scanned PDFs using pytesseract and pdf2imagein 77 of 636, across 10 files
  • Use qpdf to merge or split PDFs or large filesin 70 of 636, across 3 files
  • Use Excel formulas instead of hardcoded calculated values or Python calculationsin 68 of 636, across 13 files
  • Unpack, edit, and repack XML for existing documents or presentationsin 63 of 636, across 8 files
  • Document sources for all hardcoded valuesin 61 of 636, across 9 files
  • Write minimal, concise Python code without unnecessary commentsin 59 of 636, across 7 files
  • Run the recalculation script (recalc.py) after adding or modifying formulasin 59 of 636, across 7 files
  • Fix all identified formula errors and recalculate before finishingin 58 of 636, across 6 files
  • Format years as text stringsin 57 of 636, across 5 files

Said here and by no other author read

  • read the format standard before typesetting
  • record any local rule deviations before generation
  • derive layout from document direction and type
  • normalize content into the required JSON structure
  • inspect every rendered PNG page
  • deliver PDF with a compliance note

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 326,144. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.