PanQiWei · PanQiWei · Apr 9, 2026 · Apr 9, 2026
diff --git a/docs/en/mental-models/ch-06-causality/01-humes-fork.md b/docs/en/mental-models/ch-06-causality/01-humes-fork.md
@@ -1,6 +1,6 @@
 # Hume's Fork
 
-<div class="mm-article" markdown>
+<div class="mm-article" data-card="assets/humes-fork.png" data-card-alt="Humes Fork" markdown>
 
 For roughly two thousand years, "cause" needed no explanation.
 

diff --git a/docs/en/mental-models/ch-06-causality/02-ocean-of-correlation.md b/docs/en/mental-models/ch-06-causality/02-ocean-of-correlation.md
@@ -1,6 +1,6 @@
 # The Ocean of Correlation
 
-<div class="mm-article" markdown>
+<div class="mm-article" data-card="assets/ocean-of-correlation.png" data-card-alt="The Ocean of Correlation" markdown>
 
 Hume proved something unsettling: even humans — equipped with physical bodies and the ability to directly manipulate the world — have never actually observed causation itself. All we have is constant conjunction and habit.
 

diff --git a/docs/en/mental-models/ch-06-causality/03-the-ladder.md b/docs/en/mental-models/ch-06-causality/03-the-ladder.md
@@ -1,6 +1,6 @@
 # The Ladder
 
-<div class="mm-article" markdown>
+<div class="mm-article" data-card="assets/the-ladder.png" data-card-alt="The Ladder" markdown>
 
 How deep is the ocean of correlation? Put differently: if causal reasoning has *levels*, where does an LLM sit?
 

diff --git a/docs/en/mental-models/ch-06-causality/04-causal-discipline.md b/docs/en/mental-models/ch-06-causality/04-causal-discipline.md
@@ -1,6 +1,6 @@
 # Causal Discipline
 
-<div class="mm-article" markdown>
+<div class="mm-article" data-card="assets/causal-discipline.png" data-card-alt="Causal Discipline" markdown>
 
 The ladder made the situation clear: LLMs naturally inhabit the first rung — association. Many valuable engineering problems require second- or even third-rung reasoning.
 

diff --git a/docs/en/mental-models/ch-06-causality/05-the-cost-of-honesty.md b/docs/en/mental-models/ch-06-causality/05-the-cost-of-honesty.md
@@ -1,6 +1,6 @@
 # The Cost of Honesty
 
-<div class="mm-article" markdown>
+<div class="mm-article" data-card="assets/the-cost-of-honesty.png" data-card-alt="The Cost of Honesty" markdown>
 
 Causal discipline sounds eminently reasonable — evidence grading, temporal constraints, hypothesis verification, reasoning provenance. Who would object to "making the system more honest"?
 

diff --git a/docs/en/mental-models/ch-06-causality/06-the-dao-of-causality.md b/docs/en/mental-models/ch-06-causality/06-the-dao-of-causality.md
@@ -1,6 +1,6 @@
 # The Dao of Causality
 
-<div class="mm-article" markdown>
+<div class="mm-article" data-card="assets/the-dao-of-causality.png" data-card-alt="The Dao of Causality" markdown>
 
 Five articles in, the structure of causal discipline is clear.
 

diff --git a/docs/en/mental-models/ch-06-causality/index.md b/docs/en/mental-models/ch-06-causality/index.md
@@ -12,6 +12,10 @@ This chapter asks a structural question: when your reasoning engine learns from
 
 ---
 
+![Causality — LLM lives in an ocean of correlation, harness builds causal discipline](assets/causality-overview.png)
+
+---
+
 | # | Article | One-liner |
 |---|---------|-----------|
 | 01 | [Hume's Fork](01-humes-fork.md) | Humans themselves have never observed causation — Hume showed we rely on constant conjunction and habit, not rational proof |

diff --git a/docs/zh/mental-models/ch-06-causality/01-humes-fork.md b/docs/zh/mental-models/ch-06-causality/01-humes-fork.md
@@ -1,6 +1,6 @@
 # 休谟的叉子
 
-<div class="mm-article" markdown>
+<div class="mm-article" data-card="assets/humes-fork.png" data-card-alt="休谟的叉子" markdown>
 
 两千多年来，"因果"这个词几乎不需要解释。
 

diff --git a/docs/zh/mental-models/ch-06-causality/02-ocean-of-correlation.md b/docs/zh/mental-models/ch-06-causality/02-ocean-of-correlation.md
@@ -1,6 +1,6 @@
 # 相关性的海洋
 
-<div class="mm-article" markdown>
+<div class="mm-article" data-card="assets/ocean-of-correlation.png" data-card-alt="相关性的海洋" markdown>
 
 休谟证明了一件令人不安的事：即使是人类——拥有物理身体、能亲手推台球、能用实验干预世界的人类——也从未直接观察到因果关系本身。我们依赖的是恒常连结和习惯。
 

diff --git a/docs/zh/mental-models/ch-06-causality/03-the-ladder.md b/docs/zh/mental-models/ch-06-causality/03-the-ladder.md
@@ -1,6 +1,6 @@
 # 因果的阶梯
 
-<div class="mm-article" markdown>
+<div class="mm-article" data-card="assets/the-ladder.png" data-card-alt="因果的阶梯" markdown>
 
 相关性的海洋有多深？换一种问法：如果因果推理有"层次"，那么 LLM 停留在哪一层？
 

diff --git a/docs/zh/mental-models/ch-06-causality/04-causal-discipline.md b/docs/zh/mental-models/ch-06-causality/04-causal-discipline.md
@@ -1,6 +1,6 @@
 # 因果纪律
 
-<div class="mm-article" markdown>
+<div class="mm-article" data-card="assets/causal-discipline.png" data-card-alt="因果纪律" markdown>
 
 阶梯让我们看清了处境：LLM 天然停留在第一层——关联层。而很多有价值的工程问题需要第二层甚至第三层的推理。
 

diff --git a/docs/zh/mental-models/ch-06-causality/05-the-cost-of-honesty.md b/docs/zh/mental-models/ch-06-causality/05-the-cost-of-honesty.md
@@ -1,6 +1,6 @@
 # 诚实的代价
 
-<div class="mm-article" markdown>
+<div class="mm-article" data-card="assets/the-cost-of-honesty.png" data-card-alt="诚实的代价" markdown>
 
 因果纪律听起来很合理——证据分级、时序约束、假说验证、推理溯源。谁会反对"让系统更诚实"？
 

diff --git a/docs/zh/mental-models/ch-06-causality/06-the-dao-of-causality.md b/docs/zh/mental-models/ch-06-causality/06-the-dao-of-causality.md
@@ -1,6 +1,6 @@
 # 因果之道
 
-<div class="mm-article" markdown>
+<div class="mm-article" data-card="assets/the-dao-of-causality.png" data-card-alt="因果之道" markdown>
 
 五篇文章走下来，因果纪律的结构已经清晰了。
 

diff --git a/docs/zh/mental-models/ch-06-causality/index.md b/docs/zh/mental-models/ch-06-causality/index.md
@@ -10,6 +10,8 @@
 
 这一章追问一个结构性问题：当你的推理引擎从相关性中学习，而你的应用需要因果性的可靠性时，系统的什么地方需要补上这个差距？答案不是让模型"变得更聪明"——而是让整个系统对自己保持诚实。我们管这叫因果纪律。
 
+![因果——LLM 生活在相关性的海洋里，harness 划出因果的航道](assets/causality-overview.png)
+
 ---
 
 | # | 篇章 | 一句话 |
-Original file line number
+Diff line change
@@ Expand Up / @@ -10,6 +10,8 @@ @@
     这一章追问一个结构性问题：当你的推理引擎从相关性中学习，而你的应用需要因果性的可靠性时，系统的什么地方需要补上这个差距？答案不是让模型"变得更聪明"——而是让整个系统对自己保持诚实。我们管这叫因果纪律。
+    ![因果——LLM 生活在相关性的海洋里，harness 划出因果的航道](assets/causality-overview.png)
     ---
     | # | 篇章 | 一句话 |
@@ Expand Down @@