docs: replace README demo video with contextFlow diagram and fix design-story image path

Author: Zehee

README.md

@@ -211,9 +211,11 @@ Agent: [calls mcp__kimi-memory__tag_theme] theme=cache-design
 
 ## See It in Action
 
-The recording below is a real Kimi Code CLI session using `kimi-memory`. The user asks for two cross-session summaries — first the evolution history of the MCP memory server itself, then the evolution history of the E2E testing tools. The agent retrieves related memories and conversation turns, then synthesizes structured answers.
+The diagram below shows how `kimi-memory` sees conversation history: turns are vertical bars on a timeline, clusters group related turns, sessions are horizontal containers, and themes connect relevant turns across sessions.
 
-<video src="./assets/exp.mp4" controls width="100%"></video>
+![Turns, clusters, sessions, and theme tracing over time](./assets/contextFlow.svg)
+
+The same idea applies to the video demo: the user asks for cross-session summaries, and the agent retrieves related turns and memories along these theme lines.
 
 ## Storage Layout
 

README.zh-CN.md

@@ -211,9 +211,11 @@ Agent：[调用 mcp__kimi-memory__tag_theme] theme=cache-design
 
 ## 实际效果
 
-下面是一段真实的 Kimi Code CLI 会话录屏，使用 `kimi-memory`。用户先后要求总结 MCP 记忆服务器的演进历史、以及 E2E 测试工具的演进历史；Agent 跨会话召回相关记忆与对话 turns，并生成结构化总结。
+下图展示了 `kimi-memory` 如何看待对话历史：竖条是时间轴上的 turn，粗横线是簇，灰框是 session，彩色括号把跨 session 的相关 turns / clusters 串联成主题线。
 
-<video src="./assets/exp.mp4" controls width="100%"></video>
+![时间轴上的 turns、簇、session 与主题追溯](./assets/contextFlow.svg)
+
+同样的思路也体现在实际演示中：用户要求跨会话总结时，Agent 会沿着这些主题线召回相关的 turns 和记忆。
 
 ## 存储布局
 

assets/exp.mp4

@@ -147,7 +147,7 @@ wire.jsonl  ──►  refined/refined.sqlite  ──►  search_context / trace
 
 想象一条水平向右延伸的时间轴。上面的每一个小圆柱都是一次对话轮（turn）：
 
-![时间轴上的 turns、簇、session 与主题追溯](./assets/contextFlow.svg)
+![时间轴上的 turns、簇、session 与主题追溯](../assets/contextFlow.svg)
 
 图中元素：