feat(feishu): resume last session on /feishu start; force-fallback CardKit; rebrand to Easy Code

- feishu: persist & restore AI client history across turns so /feishu start resumes the last conversation; fix SessionManager not registering feishu sessions (auto-create metadata.json, derive UI history for getLastActiveSession)

- feishu: short-circuit CardKit 2.0 to legacy card via isCardKitV2Enabled() (unstable in prod); opt-in with EASYCODE_FEISHU_CARDKIT_V2=1

- feishu: card titles & /status version line DeepV Code -> Easy Code

- rebrand: license headers, .md docs, build/webview logs and core system prompts DeepV Code -> Easy Code (GitHub repo URLs kept)

- tests: add session round-trip / derive / short-circuit coverage; sync init prompt assertions

Author: JerryliDe

.llm-wiki/index.md

@@ -1,6 +1,6 @@
 # LLM Wiki Index
 
-> Auto-maintained by DeepV Code. Do not edit manually.
+> Auto-maintained by Easy Code. Do not edit manually.
 
 ## Sources
 

.llm-wiki/log.md

@@ -43,7 +43,7 @@
 ## [2026-05-29] feat | Release Process Guide
 
 ### Pages Added
-- `wiki/release-process.md` — 详细记录了 DeepV Code 项目的规范化发布流程和核心校验规则。
+- `wiki/release-process.md` — 详细记录了 Easy Code 项目的规范化发布流程和核心校验规则。
   - 核心限制：已推送的提交禁止 Amend 机制。
   - 核心红线：`package.json` 中的版本号必须严格低于 release tag 里的版本号，否则 CI/CD 编译发布必将失败。
   - 检索最新 Tag：使用 `git tag -l "cli-release-v*" --sort=-v:refname` 查询最新 tag 并以此计算递增下一个版本号。

.llm-wiki/wiki/GeminiClient.md

@@ -11,7 +11,7 @@ sources: [raw/02-core-module.md, raw/01-architecture.md]
 
 ## Overview
 
-`GeminiClient` 是 DeepV Code 的核心编排器，管理聊天会话、上下文压缩、循环检测和 token 限制。每个会话最多 100 轮对话。
+`GeminiClient` 是 Easy Code 的核心编排器，管理聊天会话、上下文压缩、循环检测和 token 限制。每个会话最多 100 轮对话。
 
 ## Key Responsibilities
 

.llm-wiki/wiki/ProxyAuthManager.md

@@ -11,7 +11,7 @@ sources: [raw/01-architecture.md, raw/02-core-module.md]
 
 ## Overview
 
-`ProxyAuthManager` 以单例模式管理通过 DeepV Lab 服务器的 JWT 认证。是 DeepV Code 的主要认证机制（proxy-first auth）。
+`ProxyAuthManager` 以单例模式管理通过 DeepV Lab 服务器的 JWT 认证。是 Easy Code 的主要认证机制（proxy-first auth）。
 
 ## Pattern
 

.llm-wiki/wiki/adaptive-thinking.md

@@ -7,7 +7,7 @@ sources: [packages/core/src/core/customModelAdapter.ts, packages/cli/src/ui/comm
 
 # Adaptive Thinking Mode & Reasoning
 
-DeepV Code features an advanced Adaptive Thinking Mode that provides a unified, highly optimized interface for managing LLM reasoning capabilities (such as Anthropic's Claude Thinking Budget, OpenAI's reasoning effort, and Google's Gemini thinking budget parameters).
+Easy Code features an advanced Adaptive Thinking Mode that provides a unified, highly optimized interface for managing LLM reasoning capabilities (such as Anthropic's Claude Thinking Budget, OpenAI's reasoning effort, and Google's Gemini thinking budget parameters).
 
 ## The Core Concept
 
@@ -16,7 +16,7 @@ Modern frontier models support "thinking/reasoning" before outputting their fina
 - **Google Gemini**: Uses `thinking_budget` configuration.
 - **Anthropic Claude**: Uses `thinking` with explicit `max_tokens` budgets.
 
-DeepV Code maps these distinct specifications into a unified **Thinking Effort Selector** that works across all supported models.
+Easy Code maps these distinct specifications into a unified **Thinking Effort Selector** that works across all supported models.
 
 ```
 User Selection (Thinking / Effort level)
@@ -27,7 +27,7 @@ User Selection (Thinking / Effort level)
 
 ## UI & UX Implementations
 
-Reasoning tokens can bloat the chat screen. DeepV Code renders reasoning beautifully:
+Reasoning tokens can bloat the chat screen. Easy Code renders reasoning beautifully:
 1. **Terminal / CLI UI**:
    - Displays a custom `🧠` Brain icon in front of the active model's name in the footer.
    - Collapses live thinking thoughts into an elegant, styled inline markdown block (`ReasoningDisplay.tsx`).

.llm-wiki/wiki/adding-builtin-tool-checklist.md

@@ -52,7 +52,7 @@ Why this breaks:
 
 ```ts
 const DESC =
-  'Returns the current wall-clock local time of the machine running DeepV Code.\n\n' +
+  'Returns the current wall-clock local time of the machine running Easy Code.\n\n' +
   'Use this tool to:\n' +
   '- Determine the current real-world time.\n' +
   '- Record a "start time" at the beginning of a long task.';

.llm-wiki/wiki/background-tasks.md

@@ -7,11 +7,11 @@ sources: [packages/core/src/services/backgroundTaskManager.ts, packages/cli/src/
 
 # Background Task Management
 
-The Background Task system allows DeepV Code to run persistent, long-running shell commands (such as local servers or compilation watchers) in the background while keeping the primary terminal chat interactive and responsive.
+The Background Task system allows Easy Code to run persistent, long-running shell commands (such as local servers or compilation watchers) in the background while keeping the primary terminal chat interactive and responsive.
 
 ## Why Background Tasks Exist
 
-In traditional AI CLI architectures, running a command like `npm run dev` blocks the CLI thread indefinitely. DeepV Code introduced background tasks to:
+In traditional AI CLI architectures, running a command like `npm run dev` blocks the CLI thread indefinitely. Easy Code introduced background tasks to:
 1. **Prevent Blocking**: Launch persistent server commands and allow the AI and user to continue chatting instantly.
 2. **Resource Observability**: Track output streams, exit codes, and process IDs (PIDs) directly from the terminal UI.
 3. **Control**: Empower the AI agent to list active background tasks and terminate them as needed without requiring the user to exit.

.llm-wiki/wiki/build-system.md

@@ -11,7 +11,7 @@ sources: [raw/06-build-and-scripts.md]
 
 ## Overview
 
-DeepV Code 使用两条构建管道：NPM 发布路径使用 `tsc` 逐包编译到 `dist/`，Bundle 路径使用 `esbuild` 打包为单一 `bundle/dvcode.js` 文件。
+Easy Code 使用两条构建管道：NPM 发布路径使用 `tsc` 逐包编译到 `dist/`，Bundle 路径使用 `esbuild` 打包为单一 `bundle/dvcode.js` 文件。
 
 ## Pipelines
 

.llm-wiki/wiki/cli-module.md

@@ -7,11 +7,11 @@ sources: [raw/01-architecture.md, raw/03-cli-module.md]
 
 # CLI Module
 
-> `packages/cli` — The terminal frontend of DeepV Code.
+> `packages/cli` — The terminal frontend of Easy Code.
 
 ## Overview
 
-CLI 是 DeepV Code 的终端前端包（NPM 名 `deepv-code-cli`），使用 Ink/React 构建交互式终端 UI，提供流式 AI 响应、50+ 斜杠命令、主题系统和工具确认界面。
+CLI 是 Easy Code 的终端前端包（NPM 名 `deepv-code-cli`），使用 Ink/React 构建交互式终端 UI，提供流式 AI 响应、50+ 斜杠命令、主题系统和工具确认界面。
 
 ## Package Info
 

.llm-wiki/wiki/context-compression.md

@@ -7,7 +7,7 @@ sources: [packages/core/src/services/compressionService.ts, packages/core/src/se
 
 # Context Compression & Restoration
 
-Context Compression is DeepV Code's automated, high-reliability memory pruning mechanism. It prevents large LLM sessions from exceeding their Context Window limit while preserving crucial state and reasoning context.
+Context Compression is Easy Code's automated, high-reliability memory pruning mechanism. It prevents large LLM sessions from exceeding their Context Window limit while preserving crucial state and reasoning context.
 
 ## The Challenge of Context Bloat
 
@@ -18,7 +18,7 @@ As the AI agent calls tools, processes code files, and receives linter messages,
 
 ## Architectural Mechanisms
 
-DeepV Code features a dual-layer compression pipeline consisting of **CompressionService** and **MicroCompactService**:
+Easy Code features a dual-layer compression pipeline consisting of **CompressionService** and **MicroCompactService**:
 
 ```
 Token Threshold Tripped (or >6m) → MicroCompactService