Back to Details

inquiring

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese
Interview the user until you share a reading of a question — what the evidence supports, not what to do next. Map this as a claim tree: every claim branches into the claims that hang off it (operationalizations, measurements, competing explanations, confidence).
Work the tree in rounds. The frontier is every claim whose prerequisites are already settled — the questions you can ask now without guessing at answers you haven't heard yet. Fill what you can from evidence first; ask the user only what the evidence cannot settle. Then wait.
Finding evidence is your job, never the user's. Dispatch a 
/research
agent (or equivalent lookup) for a large primary-source hunt — don't block the rest of the frontier on it. The reading is the user's: put each unsettled interpretation to them and wait. Stress-test operationalizations, sample bias, and leaps from chart to conclusion. If the user's prior survives the evidence, record it as settled with that evidence.
If a decision appears ("what should we do?"), name it and hand it to 
/grilling
. Stay on claims.
与用户沟通，直到双方对问题达成一致解读——即证据支持的结论，而非下一步行动方案。将其梳理为主张树（claim tree）：每个主张都会延伸出从属主张（包括可操作化定义、衡量标准、竞争性解释、置信度等）。
分轮次处理主张树。前沿主张指所有前提已明确的主张——即无需猜测未知答案，当前即可询问的问题。首先尽可能通过证据填补信息；仅向用户询问证据无法解决的问题，之后等待回复。
寻找证据是你的职责，而非用户的。调用
/research
Agent（或类似查询工具）来检索大量原始资料——不要因这项操作而停滞其他前沿主张的处理。解读权在用户手中：将每个未确定的解释提交给用户并等待回复。对可操作化定义、样本偏差以及从图表到结论的跳跃性逻辑进行压力测试。如果用户的原有观点经得住证据验证，则将其记录为已明确的主张并附上相关证据。
若出现决策类问题（如“我们应该怎么做？”），需明确指出该问题并转交给
/grilling
处理。始终聚焦于主张本身。

Round 0 — frame

第0轮——框架搭建

Slug the question. If 
findings/<slug>.md
already exists, reload it: settled claims stay settled; ask only the current frontier.
Otherwise write 
findings/<slug>.md
before any fetch, containing: the restated question, a proposed operationalization, in-scope claims, and unknowns. Ask the frame (is this the question, is this a fair measure, any private constraints) in the round format below. Wait. Fetch starts in Round 1 after the user accepts or amends that frame.
为问题生成简短标识（slug）。若
findings/<slug>.md
已存在，则重新加载该文件：已明确的主张保持不变；仅询问当前的前沿主张。
若文件不存在，则在任何检索操作前创建
findings/<slug>.md
，内容包括：重述后的问题、拟议的可操作化定义、范围内的主张以及未知事项。按照下方的轮次格式询问框架相关问题（这是否为核心问题？衡量标准是否公平？是否存在私有约束？）。等待用户回复。在用户接受或修改框架后，方可进入第1轮开始检索。

Later rounds

后续轮次

  1. Fetch, compute, and cite against every public claim on the frontier. Parametric knowledge is a hypothesis, not a finding. Prefer primary sources and inspectable artifacts. User-supplied private data outranks a public source when they conflict — show the conflict.
  2. Update the findings file. Open a notebook only when a claim is settled by running code (see Artifacts).
  3. Curate the chat with 
    /show-me
    (see Chat). Ask the whole remaining frontier. Wait.
A claim whose answer depends on another claim still open in this round belongs to a later round.
  1. 对前沿主张中的每个公开主张进行检索、计算并引用相关依据。参数化知识属于假设，而非结论。优先选用原始资料和可核查的工件。当用户提供的私有数据与公开来源冲突时，以私有数据为准——需明确展示冲突点。
  2. 更新结论文件。仅当通过运行代码明确主张时，才打开笔记本（参见“工件”部分）。
  3. 使用
    /show-me
    整理对话内容（参见“对话”部分）。询问所有剩余的前沿主张，等待回复。
若某一主张的答案依赖于本轮中仍未明确的另一主张，则将其归入后续轮次处理。

Round format

轮次格式

Apply the 
show-me
skill to the chat: skip the preamble, smallest visual that makes the point, brief prose beside it. Point at file anchors; do not paste the findings file. If 
show-me
is missing, use the same moves (Mermaid, trees, one HTML file).
Data-shaped views are in scope when they are the smallest view: a 2-axis sketch, a tiny comparison table, a Mermaid of competing explanations, a sparkline-in-text, an HTML annotated chart. Decoration is not.
Each frontier item:
❓ **C1** - **<claim title>**: <one-line claim / operationalization>

**A.** …
**B.** …
**C.** bounded unknown / keep digging

➡️ **B** — <one line why>
One visual may sit above the letters when the options are easier to see than to read. Letter C (or the last letter) stays available as bounded unknown. If only one reading is honest, that reading is still an option, and the 
➡️
still names it.
在对话中应用
show-me
技能：跳过开场白，使用最精简的可视化呈现核心要点，旁附简短文字说明。指向文件锚点；不要粘贴结论文件内容。若
show-me
不可用，则采用类似方式（如Mermaid图表、树形图、单个HTML文件）。
仅当数据类视图是最精简的呈现方式时才使用：如双轴草图、小型对比表格、展示竞争性解释的Mermaid图表、文本内迷你折线图、带注释的HTML图表。禁止添加装饰性元素。
每个前沿主张项格式如下：
❓ **C1** - **<主张标题>**: <一行式主张/可操作化定义>

**A.** …
**B.** …
**C.** 有界未知/继续深挖

➡️ **B** — <一行理由>
当选项可视化比文字更易理解时，可在选项上方添加一个可视化元素。选项C（或最后一个选项）始终代表“有界未知”。若只有一种合理解读，该解读仍作为选项列出，且
➡️
需指向该选项。

Evidence

证据要求

A claim stays open until the findings file records all of:
  • the claim in one sentence
  • the operationalization
  • method (search, fetch, query, code)
  • sources or cells, with enough quote or output to check the claim
  • what was not searched or run
  • uncertainty / failure modes of the method
  • verdict: settled / bounded-unknown / needs you
主张将保持未明确状态，直到结论文件记录以下全部内容：
  • 用一句话表述的主张
  • 可操作化定义
  • 方法（搜索、检索、查询、代码）
  • 来源或单元格，需包含足够的引用内容或输出结果以验证主张
  • 未检索或未执行的内容
  • 方法的不确定性/失效模式
  • 结论：已明确/有界未知/需用户参与

Artifacts

工件

Match 
/research
: put files where the repo already keeps notes; if there is none, use 
findings/
and say the path.
  • findings/<slug>.md
     — canonical claim log for the session. One file; append rounds. Resume by reloading it.
  • findings/<slug>.ipynb
     — at most one per session, created on the first computational claim. Jupyter, Python 3, pandas-class tabular work, unless the repo is clearly another data stack. No notebook when nothing needs to run.
  • show-me-{description}.html
     — a round's view when Mermaid is not enough. Disposable. Do not HTML-wrap a plot the notebook already shows unless the plot needs a curated frame.
Chat is the round protocol. Files are the evidence locker.
与
/research
保持一致：将文件存放在代码库中已有的笔记目录；若不存在此类目录，则使用
findings/
并告知路径。
  • findings/<slug>.md
     — 本次会话的标准主张日志。单个文件；逐轮追加内容。重新加载该文件即可恢复会话。
  • findings/<slug>.ipynb
     — 每次会话最多一个，在首次处理计算类主张时创建。默认使用Jupyter、Python 3、pandas类表格处理，除非代码库明确使用其他数据栈。若无需运行代码，则无需创建笔记本。
  • show-me-{description}.html
     — 当Mermaid不足以呈现时，用于展示某一轮次的视图。一次性文件。除非图表需要定制框架，否则不要将笔记本中已有的图表用HTML包裹。
对话是轮次的交互协议，文件是证据存储库。

Done

结束条件

The frontier is empty: every in-scope claim is settled or explicitly bounded-unknown, each with documentary evidence. Stop. Wait for the user to confirm the shared reading. Do not start grilling a decision off the findings.
当前沿主张为空时：所有范围内的主张均已明确或被标记为有界未知，且每个主张都附有书面证据。此时停止操作，等待用户确认共识解读。不要基于现有结论开始处理决策类问题。