paper-evidence-board

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Paper Evidence Board

论文证据看板

Maintain the paper-facing view of the research project: which claims the paper makes, what evidence supports them, where they appear, what reviewers may attack, and what actions are needed before submission.

Use this skill when:

writing reveals that a paper claim lacks evidence
new experiment results require updating paper narrative, figures, or sections
review simulation creates experiment or writing actions
a paper draft needs claim-evidence alignment before submission
figures and tables need to be mapped to paper claims
figure/table style needs to be tracked as a writing decision, including palette, markers, symbols, typography, and venue-facing visual conventions
the user wants a live evidence board for a paper
the project needs to decide whether a gap requires more experiments, rewriting, narrowing a claim, or cutting material

This skill is more specific than

research-project-memory

: it uses project memory as the source of cross-component truth, but creates a paper-facing board for writing and review decisions.

Pair this skill with:

```
research-project-memory
```
for project-level claim/evidence/risk/action IDs and writeback
```
conference-writing-adapter
```
when the board shows structural or paragraph-level writing changes
```
experiment-design-planner
```
when the board exposes missing evidence requiring new experiments
```
result-diagnosis
```
when new evidence weakens or complicates a claim
```
paper-reviewer-simulator
```
when the board should be stress-tested from a reviewer perspective
```
figure-results-review
```
when figures/tables need claim-support, caption, statistical, or visual-style review
```
citation-coverage-audit
```
when novelty or related-work gaps appear

维护研究项目的论文视角视图：论文提出了哪些论点，有哪些证据支撑这些论点，论点出现在何处，评审人员可能攻击的点，以及提交前需要完成的行动。

在以下场景中使用该技能：

写作过程中发现某一论文论点缺乏证据支撑
新的实验结果需要更新论文叙述、图表或章节
模拟评审产生了实验或写作相关的行动需求
论文草稿在提交前需要对齐论点与证据
图表需要与论文论点建立关联
需要追踪图表的风格写作决策，包括配色方案、标记、符号、排版以及符合目标会议/期刊的视觉规范
用户希望为论文创建一个实时证据看板
项目需要判断某一缺口是否需要补充实验、重写内容、缩小论点范围或删减素材

该技能比

research-project-memory

更具针对性：它将项目记忆库作为跨组件事实的来源，但会创建一个面向论文的看板，用于写作与评审决策。

搭配以下技能使用：

```
research-project-memory
```
：用于获取项目级别的论点/证据/风险/行动ID，并将内容写回项目记忆库
```
conference-writing-adapter
```
：当看板显示需要进行结构或段落级别的写作修改时
```
experiment-design-planner
```
：当看板暴露需要补充新实验的证据缺口时
```
result-diagnosis
```
：当新证据削弱或使论点复杂化时
```
paper-reviewer-simulator
```
：需要从评审人员视角对看板进行压力测试时
```
figure-results-review
```
：当图表需要进行论点支撑说明、标题、统计或视觉风格评审时
```
citation-coverage-audit
```
：当出现创新性或相关研究缺口时

Skill Directory Layout

技能目录结构

text

<installed-skill-dir>/
├── SKILL.md
└── references/
    ├── board-schema.md
    ├── evidence-gap-triage.md
    ├── paper-section-map.md
    ├── report-template.md
    ├── reviewer-risk-integration.md
    └── update-protocol.md

text

<installed-skill-dir>/
├── SKILL.md
└── references/
    ├── board-schema.md
    ├── evidence-gap-triage.md
    ├── paper-section-map.md
    ├── report-template.md
    ├── reviewer-risk-integration.md
    └── update-protocol.md

Progressive Loading

渐进式加载

Always read

references/board-schema.md

references/evidence-gap-triage.md

, and

references/update-protocol.md

Read
```
references/paper-section-map.md
```
when building a section-by-section board from a draft.
Read
```
references/reviewer-risk-integration.md
```
when using reviewer simulation, real reviews, or venue-specific risks.
Use
```
references/report-template.md
```
for substantial board reports.
Verify current venue rules or reviewer forms when venue-specific evidence expectations matter.

务必阅读

references/board-schema.md

、

references/evidence-gap-triage.md

和

references/update-protocol.md

。

当从草稿构建逐章节的看板时，阅读
```
references/paper-section-map.md
```
。
当使用模拟评审、真实评审或特定会议/期刊的风险时，阅读
```
references/reviewer-risk-integration.md
```
。
撰写完整的看板报告时，使用
```
references/report-template.md
```
。
当特定会议/期刊的证据要求至关重要时，核实当前的会议规则或评审表格。

Core Principles

核心原则

Every major paper claim should have a paper location, evidence status, and reviewer risk.
Evidence should point to source artifacts, not only prose memory.
A figure/table should have a job: support a claim, answer a reviewer question, or delimit scope.
A paper-facing figure/table should also obey the paper's visual style policy, not just display correct numbers.
Missing evidence is not a writing problem by default; it may require experiment design, result diagnosis, claim narrowing, or citation work.
Do not hide negative or weak results. Mark how they change the claim.
Keep the board actionable: each open gap should route to a next skill or action.

每一个重要的论文论点都应标注论文位置、证据状态和评审风险。
证据应指向源文件，而非仅指向文本记忆。
图表应有明确作用：支撑论点、回答评审问题或界定研究范围。
面向论文的图表还需遵循论文的视觉风格规范，而非仅展示正确的数据。
缺失证据默认不属于写作问题；它可能需要实验设计、结果诊断、缩小论点范围或引用相关文献。
不要隐藏负面或较弱的结果。标注这些结果如何改变论点。
保持看板的可操作性：每一个未填补的缺口都应指向对应的后续技能或行动。

Step 1 - Locate Paper and Project Memory

步骤1 - 定位论文与项目记忆库

Find:

paper root:
```
paper/
```
, current directory, or user-provided path
project root, if different
```
memory/claim-board.md
```
```
memory/evidence-board.md
```
```
memory/risk-board.md
```
```
memory/action-board.md
```
```
paper/.agent/paper-status.md
```
paper sources such as
```
main.tex
```
,
```
paper.tex
```
,
```
sections/*.tex
```
, figures, tables, and appendix

If project memory is missing, still build a paper board from the draft, but recommend initializing

research-project-memory

查找：

论文根目录：
```
paper/
```
、当前目录或用户提供的路径
项目根目录（若与论文根目录不同）
```
memory/claim-board.md
```
```
memory/evidence-board.md
```
```
memory/risk-board.md
```
```
memory/action-board.md
```
```
paper/.agent/paper-status.md
```
论文源文件，如
```
main.tex
```
、
```
paper.tex
```
、
```
sections/*.tex
```
、图表和附录

若项目记忆库缺失，仍可从草稿构建论文看板，但建议初始化

research-project-memory

。

Step 2 - Extract Paper Claims

步骤2 - 提取论文论点

Read the draft or provided notes and extract:

title and abstract claims
introduction contribution bullets
method claims
theory claims
experiment claims
related-work novelty boundaries
limitation and scope claims

Assign or reuse

CLM-###

IDs. For each claim, record:

exact paper location
claim type
strength
current wording risk
evidence expected
evidence available

Do not treat aspirational contribution language as proven evidence.

阅读草稿或提供的笔记，提取：

标题和摘要中的论点
引言部分的贡献要点
方法论点
理论论点
实验论点
相关研究的创新性边界
局限性与范围论点

分配或复用

CLM-###

ID。针对每个论点，记录：

确切的论文位置
论点类型
论点强度
当前表述的风险
预期的证据
已有的证据

不要将理想化的贡献表述视为已证实的证据。

Step 3 - Map Evidence to Claims

步骤3 - 关联证据与论点

Read

references/board-schema.md

For every evidence item:

identify source path, run ID, table, figure, theorem, citation, or review text
link to claim IDs
mark status: planned, running, observed, reported, stale, contradicted, or cut
mark certainty: observed, user-stated, inferred, stale, or needs-verification
state limitations

If evidence is only in a daily log or chat note, mark it accordingly and create an action to verify source artifacts.

阅读

references/board-schema.md

。

针对每一项证据：

确定源文件路径、运行ID、表格、图表、定理、引用或评审文本
关联对应的论点ID
标记状态：计划中、进行中、已观测、已报告、过时、矛盾或已删减
标记确定性：已观测、用户陈述、推断、过时或需验证
说明局限性

若证据仅存在于日常日志或聊天记录中，需相应标记并创建行动以验证源文件。

Step 4 - Build Section and Figure/Table Map

步骤4 - 构建章节与图表映射

Read

references/paper-section-map.md

Create:

section map: section -> claims -> evidence -> risks -> actions
figure/table map: figure/table -> evidence -> claim -> required caption message -> stale status
visual style map: palette, method-to-marker mapping, typography, symbols, figure sizing, and table conventions
appendix map when relevant

Identify:

claims repeated in multiple places with inconsistent strength
figures without a claim
claims without a figure/table/proof/citation when one is expected
stale figures after new results
inconsistent colors, symbols, method names, font sizes, or figure sizes across paper visuals
result tables not discussed in prose

阅读

references/paper-section-map.md

。

创建：

章节映射：章节 -> 论点 -> 证据 -> 风险 -> 行动
图表映射：图表 -> 证据 -> 论点 -> 所需标题说明 -> 过时状态
视觉风格映射：配色方案、方法与标记的对应关系、排版、符号、图表尺寸以及表格规范
附录映射（若相关）

识别：

在多个位置重复出现且强度不一致的论点
未关联任何论点的图表
预期需要图表/证明/引用但未提供的论点
新结果产生后过时的图表
论文视觉元素中颜色、符号、方法名称、字体大小或图表尺寸不一致的情况
文本中未提及的结果表格

Step 5 - Triage Evidence Gaps

步骤5 - 分类处理证据缺口

Read

references/evidence-gap-triage.md

For each gap, decide:

```
new-experiment
```
: needs
```
experiment-design-planner
```
```
result-diagnosis
```
: existing result is ambiguous or contradictory
```
rewrite
```
: evidence exists but prose overclaims or hides it
```
narrow-claim
```
: evidence supports a smaller claim
```
citation-work
```
: claim needs missing related work or attribution
```
cut
```
: claim is not worth supporting
```
accept-risk
```
: limitation can be stated rather than fixed

Every high-risk gap should create an action.

阅读

references/evidence-gap-triage.md

。

针对每个缺口，决定：

```
new-experiment
```
：需要使用
```
experiment-design-planner
```
```
result-diagnosis
```
：现有结果模糊或矛盾
```
rewrite
```
：证据存在但文本表述夸大或隐瞒
```
narrow-claim
```
：证据仅支持更小范围的论点
```
citation-work
```
：论点需要补充相关研究或引用
```
cut
```
：论点不值得花费精力支撑
```
accept-risk
```
：局限性可直接说明无需修正

每个高风险缺口都应生成对应的行动项。

Step 6 - Integrate Reviewer Risks

步骤6 - 整合评审风险

Read

references/reviewer-risk-integration.md

Use:

simulated reviewer risks from
```
paper-reviewer-simulator
```
citation risks from
```
citation-coverage-audit
```
real review risks from
```
rebuttal-strategist
```
venue-specific expectations when known

Map each risk to:

claim
paper location
evidence gap
fix type: experiment, analysis, proof, citation, rewrite, figure/table, limitation, or rebuttal
priority

阅读

references/reviewer-risk-integration.md

。

使用以下来源的风险信息：

```
paper-reviewer-simulator
```
生成的模拟评审风险
```
citation-coverage-audit
```
生成的引用风险
```
rebuttal-strategist
```
提供的真实评审风险
已知的特定会议/期刊的预期要求

将每个风险关联至：

论点
论文位置
证据缺口
修正类型：实验、分析、证明、引用、重写、图表、局限性说明或回复评审
优先级

Step 7 - Produce Board and Actions

步骤7 - 生成看板与行动项

Use

references/report-template.md

for full output.

If saving to a project and no path is given, use:

text

paper/.agent/paper-evidence-board.md

Output:

paper snapshot
claim-evidence matrix
section map
figure/table map
visual style map
evidence gaps
reviewer-risk map
action plan
project-memory writeback

撰写完整输出时使用

references/report-template.md

。

若需保存至项目且未指定路径，使用：

text

paper/.agent/paper-evidence-board.md

输出内容：

论文快照
论点-证据矩阵
章节映射
图表映射
视觉风格映射
证据缺口
评审风险映射
行动计划
项目记忆库写回内容

Step 8 - Write Back to Project Memory

步骤8 - 写回项目记忆库

If the project uses

research-project-memory

, update:

```
memory/claim-board.md
```
: claim status, wording, paper locations, and weakened/cut claims
```
memory/evidence-board.md
```
: evidence source paths, figure/table mappings, stale status, and limitations
```
memory/risk-board.md
```
: evidence-gap, writing, novelty, baseline, and reviewer risks
```
memory/action-board.md
```
: experiment, diagnosis, rewrite, citation, figure/table, and review actions
```
paper/.agent/paper-status.md
```
: current paper section state
```
paper/.agent/visual-style.md
```
: paper-facing figure/table style policy when visual conventions are material
```
paper/.agent/paper-evidence-board.md
```
: the paper-facing board

Do not duplicate long experiment reports. Link to them.

若项目使用

research-project-memory

，更新以下内容：

```
memory/claim-board.md
```
：论点状态、表述、论文位置以及已弱化/删减的论点
```
memory/evidence-board.md
```
：证据源路径、图表关联、过时状态以及局限性
```
memory/risk-board.md
```
：证据缺口、写作、创新性、基线以及评审风险
```
memory/action-board.md
```
：实验、诊断、重写、引用、图表以及评审相关行动
```
paper/.agent/paper-status.md
```
：当前论文章节状态
```
paper/.agent/visual-style.md
```
：当视觉规范至关重要时，面向论文的图表风格规范
```
paper/.agent/paper-evidence-board.md
```
：面向论文的看板

不要复制冗长的实验报告，只需链接至报告即可。

Final Sanity Check

最终检查

Before finalizing:

every major claim has a location and evidence status
unsupported claims are marked as planned, narrowed, or cut
every main figure/table has a claim-facing job
paper-facing figures/tables share consistent colors, markers, symbols, typography, sizing, and notation
stale evidence is marked
reviewer risks link to actions
next actions are routed to the right skill
project memory is updated when present

完成前需确认：

每一个重要论点都标注了位置和证据状态
无支撑的论点已标记为计划中、范围缩小或已删减
每一个主要图表都有明确的论点支撑作用
面向论文的图表在颜色、标记、符号、排版、尺寸和符号表示上保持一致
过时证据已标记
评审风险已关联至行动项
后续行动已分配至对应的技能
项目记忆库已更新（若存在）