Agent Skills
› mohitagw15856/pm-claude-skills
› thesis-outline
thesis-outline
GitHub构建以论点为核心的论文大纲，明确每章职能与主张。输出包括核心论点、章节地图、证据分配及考官风险登记册，通过模拟审查识别薄弱环节，确保逻辑连贯并避免无效结构。
触发场景
用户要求构建学位论文或博士论文大纲
需要规划毕业设计或研究项目的章节结构
希望将研究成果组织为具有论点的章节体系
安装
npx skills add mohitagw15856/pm-claude-skills --skill thesis-outline -g -y
SKILL.md
Frontmatter
{
"name": "thesis-outline",
"homepage": "https:\/\/mohitagw15856.github.io\/pm-claude-skills\/skill\/thesis-outline.html",
"metadata": {
"openclaw": {
"emoji": "🧠"
}
},
"description": "Build a defensible thesis or dissertation outline — argument-first structure, chapter by chapter, with the through-line visible. Use when asked to outline my thesis, structure my dissertation, plan my capstone, or organize my research into chapters. Produces a full outline: the one-sentence thesis, chapter map with each chapter's job and claim, evidence allocation, and the risk register of weak links an examiner would probe."
}
Thesis Outline Skill
A thesis dies two ways: chapters that are containers instead of arguments, and a through-line only the author can see. This skill builds the outline argument-first — every chapter has a job and a claim — and then reads it like an examiner, flagging the weak links before the defense does.
What This Skill Produces
- The thesis sentence — one sentence; if it takes three, the work isn't scoped yet
- The chapter map — per chapter: its job in the argument, its central claim, its evidence
- Evidence allocation — which data/sources carry which chapter, exposing overloaded and starving chapters
- The examiner's risk register — the 3–5 weakest links, each with a shoring-up plan
Required Inputs
Ask for these if not provided:
- The research question and (draft) answer — even rough; the outline is built around the answer
- What exists already — data collected, chapters drafted, the literature review's gap statement
- Program constraints — length, chapter conventions of the field, deadline
- The supervisor's known opinions (optional but valuable — outlines that fight the supervisor lose slowly)
Framework
- Answer-first: the outline is the argument for the thesis sentence; chapters exist to carry parts of that argument. "Chapter 4: Findings" is a container; "Chapter 4: the survey data shows X, which establishes premise 2" is a chapter.
- The through-line test: the final paragraph of each chapter should hand off to the next ("having shown X, the question becomes Y"). Write those handoffs in the outline itself.
- Evidence budgeting: map every dataset/source cluster to the claims it supports. One chapter hoarding all evidence while another argues from air is the most common structural failure.
- Examiner simulation: for each chapter ask, "what would a skeptical expert probe?" — method validity, alternative explanations, scope creep, the gap between claim and evidence.
- Scope honesty: everything in the outline that doesn't serve the thesis sentence gets a cut-or-justify note. Theses are late because outlines are ambitious.
Output Format
Thesis Outline: [working title]
Thesis sentence: … The argument in four moves: 1)… 2)… 3)… 4)…
Chapter map
| Ch | Title | Job in the argument | Central claim | Evidence carried |
|---|
Handoffs
[The one-sentence bridge ending each chapter]
Examiner's risk register
| Weak link | Where | The probe they'd make | Shoring-up plan |
|---|
Cut-or-justify
[Material that doesn't serve the thesis sentence, with a recommendation each]
Quality Checks
- The thesis fits in one sentence
- Every chapter has both a job and a claim — no container chapters
- Handoffs are written, making the through-line inspectable
- Evidence allocation shows no starving chapter
- The risk register names real probes with real shoring plans
Anti-Patterns
- Do not structure by activity ("what I read, what I did, what I found") — structure by argument
- Do not let chapter 2 be "everything I know about the field" — the literature review earns its place by establishing the gap, nothing more
- Do not allocate a claim no evidence supports — flag it as a collection gap now, not a defense surprise later
- Do not outline the ideal thesis — outline the one finishable with the data and months that exist
- Do not hide the weakest link — it's the first thing the committee finds; better it's the best-defended
版本历史
- 54fad50 当前 2026-07-19 12:36
