Agent SkillsNeverSight/learn-skills.dev › paper-draft-consistency-editor

paper-draft-consistency-editor

GitHub

用于编辑机器学习/AI论文草稿的内部一致性。检查并修正标题、摘要、方法、结果等部分间的声明、术语、图表及数据对齐问题,确保全文逻辑连贯且符合写作规范,而非进行创造性写作或模拟审稿。

data/skills-md/a-green-hand-jack/ml-research-skills/paper-draft-consistency-editor/SKILL.md NeverSight/learn-skills.dev

Trigger Scenarios

需要检查论文各章节(如引言、方法、结论)之间的逻辑一致性 发现术语、指标名称或图表引用存在前后不一致 需要生成论文一致性审查报告或直接修复轻微的文风/格式漂移

Install

npx skills add NeverSight/learn-skills.dev --skill paper-draft-consistency-editor -g -y
More Options

Non-standard path

npx skills add https://github.com/NeverSight/learn-skills.dev/tree/main/data/skills-md/a-green-hand-jack/ml-research-skills/paper-draft-consistency-editor -g -y

Use without installing

npx skills use NeverSight/learn-skills.dev@paper-draft-consistency-editor

指定 Agent (Claude Code)

npx skills add NeverSight/learn-skills.dev --skill paper-draft-consistency-editor -a claude-code -g -y

安装 repo 全部 skill

npx skills add NeverSight/learn-skills.dev --all -g -y

预览 repo 内 skill

npx skills add NeverSight/learn-skills.dev --list

SKILL.md

Frontmatter
{
    "name": "paper-draft-consistency-editor",
    "description": "Edit ML\/AI paper drafts for internal consistency. Use after sections exist to align claims, terminology, figures, tables, captions, limitations, and conclusion.",
    "allowed-tools": "Read, Write, Edit, Bash, Glob",
    "argument-hint": "[paper-dir] [--mode audit|edit|report] [--scope full|claims|terms|figures|results]"
}

Paper Draft Consistency Editor

Check and fix whether a paper draft is internally coherent. This skill is an editor pass, not a reviewer simulation and not a first-draft writer. It makes the title, abstract, introduction, methods, results, figures, tables, captions, related work, limitations, and conclusion agree with the paper's active claims and evidence.

Use this skill for:

  • title/abstract/intro/conclusion consistency
  • contribution bullet to experiment mapping
  • claim strength drift across sections
  • terminology, method name, dataset name, baseline name, metric name, and notation consistency
  • figure/table/caption/main-text alignment
  • result prose that no longer matches updated numbers
  • unresolved PROVISIONAL-RESULT or PR-### placeholders
  • checking whether the draft follows paper/.agent/writing-contract.md
  • producing a consistency report or directly editing narrow inconsistencies

Do not use this skill to decide the paper's positioning from scratch. Use paper-positioning-planner for that. Do not use it to write major new prose. Use paper-writing-memory-manager to record stale locations, dependency conflicts, and open writing threads found during consistency checks. Use paper-writing-assistant after consistency issues are identified. Do not simulate hostile reviewers; use paper-reviewer-simulator for acceptance-risk critique.

Skill Directory Layout

<installed-skill-dir>/
├── SKILL.md
├── references/
│   ├── consistency-rules.md
│   └── edit-protocol.md
└── templates/
    └── consistency-report.md

Progressive Loading

  • Always read references/consistency-rules.md.
  • Read references/edit-protocol.md before editing paper source.
  • Use templates/consistency-report.md for substantial reports.
  • Read paper/.agent/writing-contract.md when present.
  • Read paper/.agent/writing-memory/, paper/.agent/paper-evidence-board.md, paper/.agent/provisional-results.md, and root memory/claim-board.md / memory/evidence-board.md when present.
  • Read figure/table maps or visual-style files when checking figures and tables.

Core Principles

  • Preserve the selected paper story. Do not silently reposition the paper.
  • Distinguish inconsistency from scientific weakness. The goal is coherence, not acceptance-risk scoring.
  • Prefer the current writing contract and evidence board over local inference.
  • Fix low-risk naming, reference, and wording drift directly when requested.
  • For claim or evidence drift, report the issue and propose the smallest safe edit.
  • Do not strengthen claims to resolve inconsistencies. Weaken, narrow, or route to missing evidence instead.
  • Preserve LaTeX commands, labels, citations, math, macros, comments, and formatting unless the edit directly concerns them.

Step 1 - Locate Draft and Contract

Find:

  • paper root: paper/, current directory, or user-provided path
  • paper/.agent/writing-contract.md or .agent/writing-contract.md
  • paper/.agent/paper-evidence-board.md
  • paper/.agent/provisional-results.md
  • source files: main.tex, paper.tex, sections/*.tex, appendix, figure wrappers, table files
  • root memory: memory/claim-board.md, memory/evidence-board.md, memory/action-board.md

If no writing contract exists, still run the consistency pass from the draft, but recommend paper-writing-contract-planner.

Step 2 - Choose Mode and Scope

Modes:

  • audit: produce a report with issues and proposed fixes
  • edit: make narrow source edits for clear inconsistencies
  • report: save a report under paper/.agent/draft-consistency-report.md

Scopes:

  • full: all checks
  • claims: title/abstract/intro/result/conclusion claim alignment
  • terms: names, notation, metrics, datasets, baselines
  • figures: figures, captions, callouts, visual jobs
  • results: tables, numbers, result prose, provisional placeholders

Default to audit and full unless the user asks for direct edits.

Step 3 - Build the Draft Story Map

Extract a compact map:

## Draft Story Map
- Title promise:
- Abstract claims:
- Intro thesis:
- Contribution bullets:
- Method claims:
- Experiment claims:
- Figure/table takeaways:
- Limitation scope:
- Conclusion claims:
- Active contract:

Mark claim IDs such as CLM-### when available. If IDs are missing, use local labels like CLM-DRAFT-001.

Step 4 - Run Consistency Checks

Read references/consistency-rules.md.

Check:

  • Does the title promise match the abstract and intro thesis?
  • Do abstract results appear in the experiments/results section?
  • Do intro contribution bullets map to experiments, figures, tables, theorem, citation, or artifact evidence?
  • Are claim strengths consistent across abstract, intro, results, conclusion, and limitations?
  • Do method names, dataset names, baseline names, metric names, symbols, and notation stay consistent?
  • Are figure/table captions aligned with main-text interpretation?
  • Are result numbers and directions consistent across tables, captions, and prose?
  • Are provisional placeholders still present?
  • Does related work define the same novelty boundary as intro contributions?
  • Do limitations narrow exactly the right claims?

Classify each issue:

  • blocking: could make the draft misleading or unsupported
  • major: materially confuses the paper story
  • minor: naming, wording, or local flow inconsistency
  • style: polish issue that does not affect meaning

Step 5 - Propose or Apply Edits

Read references/edit-protocol.md.

For each issue, choose:

  • direct-edit: safe local wording/name/reference fix
  • rewrite-request: needs paper-writing-assistant
  • contract-update: needs paper-writing-contract-planner
  • evidence-update: needs paper-evidence-board
  • writing-memory-update: needs paper-writing-memory-manager to record dependency conflicts, stale locations, or closed threads
  • experiment-action: needs experiment-design-planner
  • diagnosis: needs result-diagnosis
  • leave-note: intentional inconsistency or unresolved decision

When editing:

  • edit the smallest relevant TeX or Markdown file
  • keep citations, labels, refs, macros, and math intact
  • avoid broad paragraph rewrites unless explicitly requested
  • do not replace provisional placeholders with unverified final prose

Step 6 - Produce Report

Use templates/consistency-report.md for substantial audits.

If saving and no path is specified, use:

paper/.agent/draft-consistency-report.md

Report:

  • summary of draft story
  • issue table sorted by severity
  • direct edits made, if any
  • unresolved consistency risks
  • required contract/evidence/writing follow-ups
  • provisional placeholders found

Final Sanity Check

Before finalizing:

  • no claim was strengthened without evidence
  • source edits are narrow and traceable
  • unresolved issues are routed to the right skill
  • provisional placeholders are listed
  • the final response distinguishes edits made from recommendations

Version History

  • e0220ca Current 2026-07-05 21:35

Same Skill Collection

data/skills-md/00prabalk00/claude-skills/knowledge-base-gap-finder/SKILL.md
data/skills-md/01000001-01001110/agent-jira-skills/jira-agile/SKILL.md
data/skills-md/01000001-01001110/agent-jira-skills/jira-auth/SKILL.md
data/skills-md/01000001-01001110/agent-jira-skills/jira-issues/SKILL.md
data/skills-md/01000001-01001110/agent-jira-skills/jira-project-management/SKILL.md
data/skills-md/01000001-01001110/agent-jira-skills/jira-projects/SKILL.md
data/skills-md/01000001-01001110/agent-jira-skills/jira-safe/SKILL.md
data/skills-md/01000001-01001110/agent-jira-skills/jira-search/SKILL.md
data/skills-md/01000001-01001110/agent-jira-skills/jira-spaces/SKILL.md
data/skills-md/01000001-01001110/agent-jira-skills/jira-transitions/SKILL.md
data/skills-md/0731coderlee-sudo/wechat-publisher/wechat-publisher/SKILL.md
data/skills-md/0froq/skills/conventionalcommits/SKILL.md
data/skills-md/0froq/skills/nuxt/SKILL.md
data/skills-md/0froq/skills/oq/SKILL.md
data/skills-md/0froq/skills/pinia/SKILL.md
data/skills-md/0froq/skills/pnpm/SKILL.md
data/skills-md/0froq/skills/slidev/SKILL.md
data/skills-md/0froq/skills/tsdown/SKILL.md
data/skills-md/0froq/skills/turborepo/SKILL.md
data/skills-md/0froq/skills/unocss/SKILL.md
data/skills-md/0froq/skills/vitepress/SKILL.md
data/skills-md/0froq/skills/vitest/SKILL.md
data/skills-md/0froq/skills/vue-best-practices/SKILL.md
data/skills-md/0froq/skills/vue-router-best-practices/SKILL.md
data/skills-md/0froq/skills/vue-testing-best-practices/SKILL.md
data/skills-md/0froq/skills/vue/SKILL.md
data/skills-md/0froq/skills/vueuse-functions/SKILL.md
data/skills-md/0froq/skills/web-design-guidelines/SKILL.md
data/skills-md/0juano/agent-skills/bondterminal-x402/SKILL.md
data/skills-md/0juano/agent-skills/edgeone-pages-deploy/SKILL.md
data/skills-md/0juano/agent-skills/ley-ar/SKILL.md
data/skills-md/0juano/agent-skills/ticktick/SKILL.md
data/skills-md/0juano/agent-skills/x-image-cards/SKILL.md
data/skills-md/0juano/x-image-cards/x-image-cards/SKILL.md
data/skills-md/0x0funky/agent-sprite-forge/generate2dsprite/SKILL.md
data/skills-md/0x0funky/agent-sprite-forge/video2dsprite/SKILL.md
data/skills-md/0x2e/superpowers/brainstorming/SKILL.md
data/skills-md/0x2e/superpowers/dispatching-parallel-agents/SKILL.md
data/skills-md/0x2e/superpowers/executing-plans/SKILL.md
data/skills-md/0x2e/superpowers/finishing-a-development-branch/SKILL.md
data/skills-md/0x2e/superpowers/receiving-code-review/SKILL.md
data/skills-md/0x2e/superpowers/requesting-code-review/SKILL.md
data/skills-md/0x2e/superpowers/subagent-driven-development/SKILL.md
data/skills-md/0x2e/superpowers/systematic-debugging/SKILL.md
data/skills-md/0x2e/superpowers/test-driven-development/SKILL.md
data/skills-md/0x2e/superpowers/using-git-worktrees/SKILL.md
data/skills-md/0x2e/superpowers/using-superpowers/SKILL.md
data/skills-md/0x2e/superpowers/verification-before-completion/SKILL.md
data/skills-md/0x2e/superpowers/writing-plans/SKILL.md
data/skills-md/0x2e/superpowers/writing-skills/SKILL.md

Metadata

Files
0
Version
f4b4cf1
Hash
5517b864
Indexed
2026-07-05 21:35

Accueil - Wiki
Copyright © 2011-2026 iteam. Current version is 2.155.2. UTC+08:00, 2026-08-08 07:11
浙ICP备14020137号-1 $Carte des visiteurs$