agentleFS
the-shy123456the-shy123456/thesis-docx/SKILL.md

thesis-docx

Skill472 starsChanged 7 months agoMITView on GitHub

Create, revise, and format thesis or dissertation Word documents with strict academic formatting control. Use when an AI agent needs to generate or revise thesis content, normalize Word styles, follow a school template, fix captions or page numbers or section levels, or produce evidence-based Mermaid figures and LaTeX-formatted code listings for a thesis document.

1 other project uses a version of this.

Adopted by

Most recently taken up by

  • Other versions1 project uses a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 1

    1. gxgeek-n/wenqu-mem★ 4same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

The file

Show its rules and text

We split this skill into the rules it gives an agent

50 rules. About 0 are common, 44 are distinctive, and 6 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈44

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 6

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use this skill for thesis-oriented .docx work where content quality and format fidelity both matter.

    line 10
  2. Distinctive:

    Check Microsoft Word and COM/DOM automation first.

    line 27
  3. Distinctive:

    If Word is missing, COM is unavailable, or DOM access fails:

    line 32
  4. Distinctive:

    Stop the automation-heavy plan.

    line 33
  5. Distinctive:

    Read the user's real constraints before editing.

    line 37
  6. Distinctive:

    If the user gave formal requirements, follow them strictly.

    line 40
  7. Distinctive:

    If the user did not give requirements, do not invent school-specific standards.

    line 41
  8. Very specific:

    Prefer scripts/audit_docx_ooxml.py before large-scale formatting.

    line 48
  9. Distinctive:

    Use it to detect hidden indentation, section drift, stale REF display text, and style-ID mismatches before deciding on a repair strategy.

    line 49
  10. Distinctive:

    Create missing styles only when no matching style exists.

    line 53
  11. Distinctive:

    Keep body text, headings, figure captions, table captions, references, abstract, and appendix styles separate and consistent.

    line 54
  12. Distinctive:

    Before changing any style globally, inspect whether the document contains direct paragraph formatting or OOXML overrides that will survive …

    line 56
  13. Distinctive:

    Do not invent experimental data, system structure, entities, or results.

    line 62
  14. Distinctive:

    Do not expose agent reasoning or data-feeding context inside the thesis prose.

    line 63
  15. Distinctive:

    Write in normal thesis voice, not in prompt-engineering voice, review voice, or chain-of-thought voice.

    line 70
  16. Distinctive:

    Prefer neutral academic narration that reads like student writing, e.g.

    line 72
  17. Distinctive:

    Avoid first-order meta-explanations about how the text was generated, inferred, or assembled.

    line 74
  18. Distinctive:

    Use Mermaid for architecture diagrams, E-R diagrams, flow charts, state diagrams, and similar thesis figures.

    line 77
  19. Distinctive:

    If the materials are insufficient, refuse to fabricate the figure and ask for the missing source information.

    line 82
  20. Distinctive:

    Keep only the code relevant to the argument.

    line 85
  21. Distinctive:

    Avoid synthetic filler code written only to look complete.

    line 87
  22. Distinctive:

    Run a structure audit first: styles, sections, captions, references, page number scheme, cross-references, hidden paragraph overrides.

    line 89
  23. Distinctive:

    Only say the task is complete after the PDF-level audit passes or after you clearly state the remaining manual visual checks.

    line 92
  24. Distinctive:

    Treat styles as the single source of truth for formatting.

    line 97
  25. Distinctive:

    Prefer these logical style buckets:

    line 98
  26. Distinctive:

    If the document already has equivalent styles, map to them and normalize their font, spacing, indentation, and numbering behavior.

    line 107
  27. Distinctive:

    If direct formatting conflicts with styles, reduce the direct formatting and bring the document back under style control.

    line 109
  28. Distinctive:

    Do not globally normalize sections that the school guide does not specify.

    line 111
  29. Very specific:

    When a visual issue remains after style normalization, inspect hidden OOXML state such as firstLineChars, numbering indentation, titlePg, …

    line 116
  30. Distinctive:

    Before bulk editing, produce and internally follow a checklist like this:

    line 123
  31. Distinctive:

    Do not say "finished" merely because a structural audit looks good.

    line 132
  32. Very specific:

    Read references/failure-patterns-and-quality-gates.md before large-scale formatting work.

    line 137
  33. Distinctive:

    only then final delivery language

    line 158
  34. Distinctive:

    If page rendering cannot be verified, say so explicitly instead of implying the formatting is fully validated.

    line 160
  35. Distinctive:

    Use Mermaid when a thesis figure is needed and the structure can be traced to real source material.

    line 165
  36. Distinctive:

    Keep the diagram academically neutral and concise.

    line 167
  37. Distinctive:

    Avoid decorative labels, chatty callouts, and speculative entities.

    line 168
  38. Very specific:

    Read references/figure-and-code-rules.md before generating diagrams.

    line 170
  39. Very specific:

    Read references/thesis-voice-and-style.md before generating or rewriting body text.

    line 174
  40. Distinctive:

    do not leak the existence of source-feeding, code-ingestion, or analysis steps into the final paper text

    line 181
  41. Distinctive:

    Before finalizing thesis prose, quickly self-check:

    line 184
  42. Distinctive:

    Prefer LaTeX-oriented code presentation when the thesis includes code listings.

    line 196
  43. Distinctive:

    Keep the listing faithful to the actual code.

    line 198
  44. Distinctive:

    If the user requests a specific LaTeX package or listing style, follow it.

    line 200
  45. Distinctive:

    Read for the standard Word thesis formatting workflow.

    line 218
  46. Distinctive:

    Read before generating Mermaid figures or LaTeX code listings.

    line 220
  47. Distinctive:

    Read before generating or rewriting thesis prose.

    line 222
  48. Distinctive:

    Read before large-scale formatting or before claiming the thesis is fully checked.

    line 224
  49. Distinctive:

    Read for command examples and config file conventions.

    line 227
  50. Distinctive:

    If Word automation or PDF verification was unavailable, explicitly warn that layout fidelity was not guaranteed.

    line 241

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Thesis DOCX

Overview

Use this skill for thesis-oriented .docx work where content quality and format fidelity both matter. Prefer Microsoft Word desktop automation over WPS-like alternatives whenever the task involves batch formatting, styles, captions, pagination, tables of contents, or cross-references.

This skill is designed to avoid the most common thesis-editing failure mode: the agent makes broad formatting changes too early, introduces new layout problems, and then forces the user to catch them one by one. The default behavior should therefore be:

  1. audit first,
  2. separate explicit school requirements from unspecified formatting,
  3. fix only what is justified,
  4. re-check page by page before claiming completion.

Workflow

  1. Check Microsoft Word and COM/DOM automation first.
    • On Windows, run:
    powershell -ExecutionPolicy Bypass -File scripts/check_word_com.ps1 -Json
    • If Word is missing, COM is unavailable, or DOM access fails:
      • Stop the automation-heavy plan.
      • Tell the user to install desktop Microsoft Word.
      • Explain briefly that WPS or similar tools are likely to degrade layout fidelity for thesis formatting.
  2. Read the user's real constraints before editing.
    • Collect the thesis template, school formatting guide, screenshots, existing document, sample pages, and any explicit chapter rules.
    • If the user gave formal requirements, follow them strictly.
    • If the user did not give requirements, do not invent school-specific standards. Use conservative academic defaults and say that they are defaults.
    • Explicitly classify requirements into two buckets before editing:
      • must enforce: school guide, template, user-confirmed style rules
      • preserve current state: anything the school guide does not define
  3. Audit the DOCX / OOXML before risky bulk edits.
    • Prefer scripts/audit_docx_ooxml.py before large-scale formatting.
    • Use it to detect hidden indentation, section drift, stale REF display text, and style-ID mismatches before deciding on a repair strategy.
  4. Standardize with styles, not scattered direct formatting.
    • Reuse and repair existing styles whenever possible.
    • Create missing styles only when no matching style exists.
    • Keep body text, headings, figure captions, table captions, references, abstract, and appendix styles separate and consistent.
    • Before changing any style globally, inspect whether the document contains direct paragraph formatting or OOXML overrides that will survive the style change.
  5. Edit content only from user-provided facts.
    • Expand, polish, or reorganize thesis text only within the user's topic, evidence, codebase, notes, or source material.
    • Do not invent experimental data, system structure, entities, or results.
    • Do not expose agent reasoning or data-feeding context inside the thesis prose. Forbidden patterns include thesis-language such as:
      • "根据已有工程"
      • "根据现有代码"
      • "将任务书和代码喂给模型后"
      • "通过分析用户提供的代码"
      • any wording that sounds like an AI workflow note instead of thesis prose
    • Write in normal thesis voice, not in prompt-engineering voice, review voice, or chain-of-thought voice.
    • Prefer neutral academic narration that reads like student writing, e.g. "本文设计并实现了……", "系统采用了……", "在该模块中……"
    • Avoid first-order meta-explanations about how the text was generated, inferred, or assembled.
  6. Generate figures only when the source material is sufficient.
    • Use Mermaid for architecture diagrams, E-R diagrams, flow charts, state diagrams, and similar thesis figures.
    • Base every node, field, relation, and dependency on real materials from the user, such as code, SQL schema, API docs, project docs, or the thesis draft itself.
    • If the materials are insufficient, refuse to fabricate the figure and ask for the missing source information.
  7. Typeset code with LaTeX conventions when thesis code excerpts are needed.
    • Keep only the code relevant to the argument.
    • Preserve real identifiers from the user's code or design.
    • Avoid synthetic filler code written only to look complete.
  8. Audit before you claim completion.
    • Run a structure audit first: styles, sections, captions, references, page number scheme, cross-references, hidden paragraph overrides.
    • Then export the document to PDF from Word and review page by page.
    • Only say the task is complete after the PDF-level audit passes or after you clearly state the remaining manual visual checks.

Style Strategy

  • Treat styles as the single source of truth for formatting.
  • Prefer these logical style buckets:
    • Body Text
    • Heading 1 / Heading 2 / Heading 3
    • Figure Caption
    • Table Caption
    • References
    • Abstract
    • Keywords
    • Appendix Title
  • If the document already has equivalent styles, map to them and normalize their font, spacing, indentation, and numbering behavior.
  • If direct formatting conflicts with styles, reduce the direct formatting and bring the document back under style control.
  • Do not globally normalize sections that the school guide does not specify. Examples of high-risk overreach:
    • page header/footer redesign when the user only asked for headings
    • changing code box appearance when the task is only about references
    • normalizing table internals when the school guide does not regulate them
  • When a visual issue remains after style normalization, inspect hidden OOXML state such as firstLineChars, numbering indentation, titlePg, differentFirstPageHeaderFooter, REF field display text, and direct run formatting.

Audit-First Discipline

Before bulk editing, produce and internally follow a checklist like this:

  1. Which parts are explicitly regulated by the school guide?
  2. Which parts are user-defined house rules?
  3. Which parts are currently acceptable and must be preserved?
  4. Which problems are structural vs. visual-only?
  5. Which fixes can be done safely through styles?
  6. Which fixes require Word COM or OOXML-level patching?

Do not say "finished" merely because a structural audit looks good. For thesis work, pagination and page-level rendering are part of correctness.

High-Risk Pitfalls

Read references/failure-patterns-and-quality-gates.md before large-scale formatting work. In particular, guard against:

  • style names that look correct but map to the wrong style IDs
  • paragraph-level direct formatting that overrides the intended style
  • firstLineChars or numbering indentation creating invisible extra indents
  • TOC fields or cross-reference fields showing stale display text
  • section-level first-page settings causing missing headers or page numbers
  • Word vs. WPS differences for table row height, vertical alignment, and code box title clipping
  • punctuation normalization that accidentally rewrites DOI, URLs, code, or English references

Visual Review Rule

For school-format-sensitive thesis work, final verification should prefer this order:

  1. Word COM/DOM structural checks
  2. Word export to PDF
  3. page-by-page PDF review
  4. only then final delivery language

If page rendering cannot be verified, say so explicitly instead of implying the formatting is fully validated.

Figure Rules

  • Use Mermaid when a thesis figure is needed and the structure can be traced to real source material.
  • Keep the diagram academically neutral and concise.
  • Avoid decorative labels, chatty callouts, and speculative entities.
  • Match the user's terminology unless it conflicts with the real materials.
  • Read references/figure-and-code-rules.md before generating diagrams.

Thesis Voice Rule

Read references/thesis-voice-and-style.md before generating or rewriting body text. In particular:

  • thesis prose must sound like thesis prose, not like notes about how the AI reasoned
  • student-facing academic narration is preferred over review-style meta commentary
  • do not leak the existence of source-feeding, code-ingestion, or analysis steps into the final paper text

Before finalizing thesis prose, quickly self-check:

  1. Does the paragraph describe the system/research itself, rather than how the assistant inferred it?
  2. Does the paragraph sound like thesis narration instead of prompt notes?
  3. Does the paragraph avoid phrases like "根据现有代码", "根据已有工程", "通过分析用户提供的代码", or similar workflow-language?
  4. Would the sentence still make sense if the reader had no idea an AI helped draft it?

Code Listing Rules

  • Prefer LaTeX-oriented code presentation when the thesis includes code listings.
  • Keep the listing faithful to the actual code.
  • Trim non-essential boilerplate when it does not support the thesis argument.
  • If the user requests a specific LaTeX package or listing style, follow it.

Resource Guide

  • scripts/check_word_com.ps1
    • Detect whether Microsoft Word desktop and COM/DOM automation are available.
  • scripts/audit_docx_ooxml.py
    • Audit DOCX styles, direct indentation, section settings, numbering, and REF field behavior before making risky formatting changes.
  • scripts/normalize_word_styles.ps1
    • Batch-normalize thesis body text, Heading 1-3, figure captions, and table captions through Word COM automation.
  • scripts/export_word_pdf.ps1
    • Export the current thesis document to PDF through Word COM for page-level review.
  • scripts/render_mermaid_figure.ps1
    • Render Mermaid source into thesis-ready SVG, PNG, or PDF figure assets.
  • references/paper-format-workflow.md
    • Read for the standard Word thesis formatting workflow.
  • references/figure-and-code-rules.md
    • Read before generating Mermaid figures or LaTeX code listings.
  • references/thesis-voice-and-style.md
    • Read before generating or rewriting thesis prose.
  • references/failure-patterns-and-quality-gates.md
    • Read before large-scale formatting or before claiming the thesis is fully checked.
  • references/script-usage.md
    • Read for command examples and config file conventions.

Final Checks

  • Confirm the document is still style-driven after edits.
  • Confirm captions, numbering, page breaks, and table of contents are coherent.
  • Confirm cross-references display the current target labels, not stale field text.
  • Confirm section settings do not silently remove page headers or page numbers from first pages.
  • Confirm references are formatted according to the school rules and that punctuation normalization did not damage DOI or English references.
  • Confirm every diagram and code block is grounded in user-provided material.
  • Export to PDF and inspect every page before saying the format is complete.
  • If Word automation or PDF verification was unavailable, explicitly warn that layout fidelity was not guaranteed.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.