paper-profile

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Paper Profile

论文档案

Elicits the author's paper positioning once, stores it in

.paper-memory/profile.yml

, and lets every other skill read it so the whole toolkit behaves context-awarely instead of asking the same questions over and over. This is the "ask me about my paper" front door: a short, optioned interview, then a small validated YAML file the rest of the repo consumes.

It is a copilot, not an oracle. The profile records your stated intent (what kind of paper this is, who it's for, how bold you want to be). It does not judge whether the science is good and it never predicts acceptance.

一次性获取作者的论文定位，并将其存储在

.paper-memory/profile.yml

中，让其他所有skill都能读取该信息，从而使整个工具包实现上下文感知，无需反复询问相同问题。这是“询问我的论文相关信息”的入口：通过简短的选项式访谈生成一份经过验证的小型YAML文件，供仓库中其他工具调用。

它是一个copilot，而非预言机。该档案记录的是你明确的意图（这是一篇什么类型的论文、受众是谁、你希望多大胆）。它不会评判科研内容的优劣，也绝不会预测论文是否能被录用。

When to use

使用场景

"Set up my paper profile" / "ask me about my paper" / "what verticals?"
Starting a new paper and wanting the toolkit tuned before drafting.
"Remember my writing style / positioning across skills."
Any other skill notices
```
.paper-memory/profile.yml
```
is missing and you want to create it so that skill can personalize.
Re-run any time the positioning changes (e.g. you drop down a venue tier, or pivot from systems to empirical framing).

当你说出“设置我的论文档案”/“询问我的论文相关信息”/“属于哪些研究方向？”时
开始撰写新论文，希望在起草前调整工具包时
需要“让所有skill记住我的写作风格/定位”时
其他skill检测到
```
.paper-memory/profile.yml
```
缺失，你希望创建该档案以实现个性化功能时
当论文定位发生变化时（例如，你选择了更低层级的会议，或从系统方向转向实证框架），可重新运行本工具

Inputs

输入信息

The user's paper working directory (where they want
```
.paper-memory/
```
to live). This is the user's paper repo, not this skills repo.
The user's answers to the interview (you ask; they pick). Nothing else is required — there is no network call and no file the user must pre-create.
Optional: an existing
```
.paper-memory/profile.yml
```
to update instead of starting fresh.

用户的论文工作目录（希望存放
```
.paper-memory/
```
的位置）。这是用户的论文仓库，而非本skill仓库。
用户对访谈问题的回答（由工具提问，用户选择）。无需其他额外信息——无需网络调用，也无需用户预先创建任何文件。
可选：已存在的
```
.paper-memory/profile.yml
```
文件，用于更新而非重新创建

Process

操作流程

Locate the paper directory and the memory dir. Confirm with the user where their paper lives; the profile goes in
```
<paper-dir>/.paper-memory/
```
. If one already exists, load and show it (
```
profile_io.py show
```
) and offer to update rather than overwrite.
Show the option menu, then interview. Print the blank template so the user sees the choices, then ask through them. Get the exact vocabulary from the script so you never invent a value:
```
python3 scripts/profile_io.py schema
```
. Ask in this order, always offering the options and a one-line gloss of each (full descriptions live in references/positioning-axes.md):
- vertical / emphasis (required):
```
systems
```
  |
```
theory
```
  |
```
applied
```
  |
```
empirical
```
  |
```
survey
```
  |
```
position
```
  . "Is the heart of the paper a built artifact, a proof, a domain application, a measurement study, a synthesis, or an argument?"
- contribution_type (required):
```
method
```
  |
```
system
```
  |
```
theory
```
  |
```
dataset
```
  |
```
empirical
```
  |
```
application
```
  |
```
survey
```
  |
```
position
```
  .
- audience:
```
specialists
```
  |
```
broad-field
```
  |
```
practitioners
```
  |
```
interdisciplinary
```
  .
- venue_tier (required):
```
top
```
  |
```
specialized
```
  |
```
regional
```
  |
```
journal
```
  |
```
workshop
```
  |
```
preprint
```
  |
```
undecided
```
  ; plus any concrete
```
target_venues
```
  (e.g.
```
sigspatial-2026
```
  ).
- risk_appetite (required):
```
safe
```
  |
```
balanced
```
  |
```
ambitious
```
  . "Defend a tight incremental delta, or stake a big claim and accept polarized reviews?"
- writing_preferences:
```
person
```
  (we/I/impersonal/venue-default),
```
tone
```
  ,
```
notation
```
  (heavy/light),
```
british_spelling
```
  ; plus
```
preferred_terms
```
  and
```
avoid_terms
```
  .
- context:
```
prior_papers
```
  (paths/ids, for
```
match-style
```
  ),
```
constraints
```
  (hard deadline, must stay anonymized, no new experiments).
Offer a sensible default for each and let the user accept it. Do not force every field — only the four required ones must be set.

Write and validate. Persist the answers with one call. The script validates against the closed vocabulary and stamps the date; it refuses to write an out-of-vocabulary value (so a typo can't silently corrupt the file that every other skill trusts):

python3 scripts/profile_io.py write <paper-dir>/.paper-memory/profile.yml \
    --field vertical=systems --field contribution_type=system \
    --field venue_tier=top --field risk_appetite=ambitious \
    --field audience=broad-field \
    --field wp.person=first-person-we --field wp.tone=assertive \
    --field target_venues="sigspatial-2026, vldb-2027" \
    --field constraints="hard deadline 2026-08-01, no new experiments"

Use

--from <existing>

to update in place, repeated

--field

to set each answer, comma-separated values for list fields, and

wp.<key>=...

for the nested writing preferences. Re-validate any hand-edited file with

python3 scripts/profile_io.py validate <path>

Set up .gitignore (ask first).
```
.paper-memory/
```
is local — it holds the author's private positioning and accumulated lessons. Recommend adding
```
.paper-memory/
```
to the paper repo's
```
.gitignore
```
unless the user deliberately wants to version it (e.g. to share positioning with co-authors). State the choice; let the user decide.
Explain how downstream skills consume it. Tell the user concretely what changes now that the profile exists (see the table below and references/downstream-consumption.md). The point of the interview is that they won't be re-asked.
Mention the rest of
.paper-memory/
. This skill owns
```
profile.yml
```
. The same directory also accumulates
```
lessons.md
```
(deduped, dated lessons other skills append when they catch something, and read at start to avoid repeating advice) and
```
decisions.md
```
(venue/track/positioning decisions with rationale). See
the shared
```
.paper-memory/
```
convention
for the file formats and memory-hygiene rules. This skill does not write those two files; it just establishes the directory and explains them.

定位论文目录和记忆目录。与用户确认论文所在位置；档案将存放在
```
<paper-dir>/.paper-memory/
```
中。如果该目录已存在，则加载并显示现有档案（执行
```
profile_io.py show
```
），并提供更新选项而非直接覆盖。
展示选项菜单并进行访谈。先显示空白模板让用户了解可选内容，然后逐一提问。所有选项词汇必须严格来自脚本，不得自行创造：执行
```
python3 scripts/profile_io.py schema
```
获取词汇。按以下顺序提问，每个问题都需提供选项及一行简要说明（详细说明见references/positioning-axes.md）：
- 研究方向/侧重点（必填）：
```
systems
```
  |
```
theory
```
  |
```
applied
```
  |
```
empirical
```
  |
```
survey
```
  |
```
position
```
  。“论文的核心是构建的产物、理论证明、领域应用、测量研究、综述，还是论点？”
- 贡献类型（必填）：
```
method
```
  |
```
system
```
  |
```
theory
```
  |
```
dataset
```
  |
```
empirical
```
  |
```
application
```
  |
```
survey
```
  |
```
position
```
  。
- 受众：
```
specialists
```
  |
```
broad-field
```
  |
```
practitioners
```
  |
```
interdisciplinary
```
  。
- 会议层级（必填）：
```
top
```
  |
```
specialized
```
  |
```
regional
```
  |
```
journal
```
  |
```
workshop
```
  |
```
preprint
```
  |
```
undecided
```
  ；可附加具体的
```
target_venues
```
  （例如
```
sigspatial-2026
```
  ）。
- 风险偏好（必填）：
```
safe
```
  |
```
balanced
```
  |
```
ambitious
```
  。“是保守地论证微小的增量改进，还是提出大胆主张并接受两极分化的审稿意见？”
- 写作偏好：
```
person
```
  （we/I/非人称/会议默认）、
```
tone
```
  、
```
notation
```
  （密集/简洁）、
```
british_spelling
```
  ；可附加
```
preferred_terms
```
  和
```
avoid_terms
```
  。
- 上下文信息：
```
prior_papers
```
  （文件路径/编号，用于
```
match-style
```
  功能）、
```
constraints
```
  （严格截止日期、必须匿名、不得开展新实验）。
每个问题都提供合理默认值，用户可直接接受。无需强制填写所有字段——仅需填写四个必填字段即可。

写入并验证。通过一次调用保存答案。脚本会对照预设词汇进行验证并标记日期；拒绝写入超出词汇范围的值（避免输入错误导致其他skill依赖的文件被静默损坏）：

python3 scripts/profile_io.py write <paper-dir>/.paper-memory/profile.yml \
    --field vertical=systems --field contribution_type=system \
    --field venue_tier=top --field risk_appetite=ambitious \
    --field audience=broad-field \
    --field wp.person=first-person-we --field wp.tone=assertive \
    --field target_venues="sigspatial-2026, vldb-2027" \
    --field constraints="hard deadline 2026-08-01, no new experiments"

使用

--from <existing>

参数进行原地更新，重复使用

--field

参数设置每个答案，列表字段使用逗号分隔值，嵌套写作偏好使用

wp.<key>=...

格式。可通过

python3 scripts/profile_io.py validate <path>

重新验证手动编辑的文件。

设置.gitignore（需先询问用户）。
```
.paper-memory/
```
是本地文件——它存储了作者的私人定位和积累的经验。建议将
```
.paper-memory/
```
添加到论文仓库的
```
.gitignore
```
中，除非用户明确希望对其进行版本控制（例如，与合作者共享定位信息）。告知用户该选项，由用户自行决定。
解释下游skill如何使用该档案。向用户具体说明档案创建后会带来哪些变化（见下表及references/downstream-consumption.md）。访谈的意义在于无需反复询问相同问题。
提及
.paper-memory/
目录下的其他文件。本skill负责管理
```
profile.yml
```
。同一目录下还会积累
```
lessons.md
```
（去重并标记日期的经验，由其他skill在发现问题时追加，启动时读取以避免重复给出建议）和
```
decisions.md
```
（记录会议/赛道/定位决策及理由）。有关文件格式和记忆管理规则，请参阅
共享
```
.paper-memory/
```
约定
。本skill不会写入这两个文件，仅负责创建目录并进行说明。

How downstream skills consume the profile

下游skill如何使用该档案

Skill	Reads	Effect
`benchmark-paper`	`vertical` , `contribution_type`	Re-weights scorecard dimensions (a `theory` paper isn't penalized for a thin evaluation; a `system` paper is).
`simulate-reviewers`	`vertical` , `venue_tier` , `risk_appetite`	Calibrates reviewer personas + harshness; an `ambitious` claim at a `top` venue gets a skeptic, not a rubber stamp.
`polish-prose`	`writing_preferences` , `avoid_terms`	Tunes the de-AI-ify / register pass to the author's person, tone, spelling, and banned terms.
`match-style`	`writing_preferences` , `prior_papers` , `preferred_terms`	Seeds the target voice and terminology so alignment matches the author.
`tailor-to-venue` / `select-venue`	`venue_tier` , `target_venues` , `contribution_type`	Frames the contribution and shortlist toward the stated targets.
`write-abstract`	`vertical` , `key_claim` , `audience`	Leads with the claim the right reader cares about.

These skills should degrade gracefully: if

profile.yml

is absent they ask the user (or use venue defaults) as they do today. The profile removes the re-asking; it is never a hard dependency.

Skill	读取字段	作用
`benchmark-paper`	`vertical` , `contribution_type`	重新调整评分卡维度的权重（ `theory` 论文不会因评估内容单薄而被扣分； `system` 论文则会）。
`simulate-reviewers`	`vertical` , `venue_tier` , `risk_appetite`	校准审稿人角色及严苛程度；在顶级会议上提出 `ambitious` 主张会遇到挑剔的审稿人，而非轻易通过。
`polish-prose`	`writing_preferences` , `avoid_terms`	根据作者的人称、语气、拼写和禁用术语调整去AI化/语体适配流程。
`match-style`	`writing_preferences` , `prior_papers` , `preferred_terms`	设定目标语气和术语，使内容风格与作者保持一致。
`tailor-to-venue` / `select-venue`	`venue_tier` , `target_venues` , `contribution_type`	根据设定目标调整贡献表述和会议候选列表。
`write-abstract`	`vertical` , `key_claim` , `audience`	以目标读者关心的主张作为摘要开头。

这些skill具备优雅降级能力：如果

profile.yml

不存在，它们会像现在一样询问用户（或使用会议默认值）。该档案的作用是避免重复询问，绝非硬性依赖。

Output

输出结果

<paper-dir>/.paper-memory/profile.yml

— a small validated YAML file (schema v1). Required:

vertical

contribution_type

venue_tier

risk_appetite

. Optional everything else.

A short plain-English recap of the positioning and the concrete behavior changes in the skills the user is likely to run next.

```
<paper-dir>/.paper-memory/profile.yml
```
——一份经过验证的小型YAML文件（schema v1）。必填字段：
```
vertical
```
,
```
contribution_type
```
,
```
venue_tier
```
,
```
risk_appetite
```
。其余字段为可选。
一份简洁的英文总结，说明论文定位以及用户接下来可能使用的skill会发生哪些具体行为变化。

Adapt to your discipline

适配其他学科

The positioning axes are CS-flavored (systems/theory/empirical, conference tiers, double-blind constraints). For other fields, edit the vocabularies at the top of

scripts/profile_io.py

(e.g. add

clinical-trial

humanities-essay

verticals, swap venue tiers for journal quartiles) — the validator and emitter are data-driven, so new disciplines need new tokens, not new code. Bump

SCHEMA_VERSION

if you change required fields.

定位维度默认适用于计算机科学（系统/理论/实证、会议层级、双盲约束）。对于其他学科，可编辑

scripts/profile_io.py

顶部的词汇表（例如添加

clinical-trial

或

humanities-essay

研究方向，将会议层级替换为期刊分区）——验证器和生成器均为数据驱动，因此新学科仅需添加新词汇，无需修改代码。如果更改了必填字段，请升级

SCHEMA_VERSION

。

Guardrails

约束规则

Never invent a vocabulary value. Pull the allowed values from
```
profile_io.py schema
```
; the writer rejects anything else so the file every other skill trusts can't be silently corrupted.
Never decide for the user. Offer options and a default; the author picks their own positioning and risk appetite. Don't infer "ambitious" because the topic sounds exciting.
Local and private.
```
.paper-memory/
```
is never uploaded anywhere; always offer the
```
.gitignore
```
line and respect the user's versioning choice.
This skill is configuration, not authorship. It records intent. It does not write paper content, predict acceptance, or submit anything.
Keep this file under 500 lines;
```
references/
```
go one level deep only.

不得自行创造词汇值。从
```
profile_io.py schema
```
中获取允许的值；写入器会拒绝其他任何值，以避免其他skill依赖的文件被静默损坏。
不得替用户做决定。提供选项和默认值；由作者自行选择定位和风险偏好。不得因主题听起来有趣就推断用户偏好
```
ambitious
```
。
本地私有。
```
.paper-memory/
```
绝不会上传到任何地方；始终建议添加.gitignore规则，并尊重用户的版本控制选择。
本skill仅负责配置，不负责写作。它仅记录意图，不会撰写论文内容、预测录用结果或提交任何文件。
保持本文件不超过500行；
```
references/
```
目录仅允许一级子目录。