@kkelly-offical/kkcode 0.1.6 → 0.2.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (163) hide show
  1. package/LICENSE +674 -674
  2. package/README.md +452 -387
  3. package/package.json +50 -46
  4. package/src/agent/agent.mjs +19 -2
  5. package/src/agent/custom-agent-loader.mjs +6 -3
  6. package/src/agent/generator.mjs +2 -2
  7. package/src/agent/prompt/assistant.txt +12 -0
  8. package/src/agent/prompt/bug-hunter.txt +90 -0
  9. package/src/agent/prompt/frontend-designer.txt +58 -58
  10. package/src/agent/prompt/guide.txt +1 -1
  11. package/src/agent/prompt/longagent-blueprint-agent.txt +83 -83
  12. package/src/agent/prompt/longagent-coding-agent.txt +37 -37
  13. package/src/agent/prompt/longagent-debugging-agent.txt +46 -46
  14. package/src/agent/prompt/longagent-preview-agent.txt +63 -63
  15. package/src/command/custom-commands.mjs +2 -2
  16. package/src/commands/agent.mjs +1 -1
  17. package/src/commands/background.mjs +145 -4
  18. package/src/commands/chat.mjs +117 -76
  19. package/src/commands/config.mjs +148 -1
  20. package/src/commands/doctor.mjs +30 -6
  21. package/src/commands/init.mjs +32 -6
  22. package/src/commands/longagent.mjs +117 -0
  23. package/src/commands/mcp.mjs +275 -43
  24. package/src/commands/permission.mjs +1 -1
  25. package/src/commands/session.mjs +195 -140
  26. package/src/commands/skill.mjs +63 -0
  27. package/src/commands/theme.mjs +1 -1
  28. package/src/config/defaults.mjs +280 -260
  29. package/src/config/import-config.mjs +1 -1
  30. package/src/config/load-config.mjs +61 -4
  31. package/src/config/schema.mjs +591 -574
  32. package/src/context.mjs +4 -1
  33. package/src/core/constants.mjs +97 -91
  34. package/src/core/types.mjs +1 -1
  35. package/src/github/api.mjs +78 -78
  36. package/src/github/auth.mjs +294 -286
  37. package/src/github/flow.mjs +298 -298
  38. package/src/github/workspace.mjs +225 -212
  39. package/src/index.mjs +84 -82
  40. package/src/knowledge/frontend-aesthetics.txt +38 -38
  41. package/src/mcp/client-http.mjs +139 -141
  42. package/src/mcp/client-sse.mjs +297 -288
  43. package/src/mcp/client-stdio.mjs +534 -533
  44. package/src/mcp/constants.mjs +2 -2
  45. package/src/mcp/registry.mjs +498 -479
  46. package/src/mcp/stdio-framing.mjs +135 -133
  47. package/src/mcp/tool-result.mjs +24 -24
  48. package/src/observability/edit-diagnostics.mjs +449 -0
  49. package/src/observability/index.mjs +42 -42
  50. package/src/observability/metrics.mjs +165 -137
  51. package/src/observability/tracer.mjs +137 -137
  52. package/src/onboarding.mjs +209 -0
  53. package/src/orchestration/background-manager.mjs +567 -372
  54. package/src/orchestration/background-worker.mjs +419 -305
  55. package/src/orchestration/interruption-reason.mjs +21 -0
  56. package/src/orchestration/longagent-manager.mjs +197 -171
  57. package/src/orchestration/stage-scheduler.mjs +733 -728
  58. package/src/orchestration/subagent-router.mjs +7 -1
  59. package/src/orchestration/task-scheduler.mjs +219 -7
  60. package/src/permission/engine.mjs +1 -1
  61. package/src/permission/exec-policy.mjs +370 -370
  62. package/src/permission/file-edit-policy.mjs +108 -0
  63. package/src/permission/prompt.mjs +1 -1
  64. package/src/permission/rules.mjs +116 -7
  65. package/src/plugin/builtin-hooks/post-edit-format.mjs +2 -1
  66. package/src/plugin/builtin-hooks/post-edit-typecheck.mjs +104 -40
  67. package/src/plugin/hook-bus.mjs +19 -5
  68. package/src/plugin/manifest-loader.mjs +222 -0
  69. package/src/provider/anthropic.mjs +396 -390
  70. package/src/provider/ollama.mjs +7 -1
  71. package/src/provider/openai.mjs +382 -340
  72. package/src/provider/retry-policy.mjs +74 -68
  73. package/src/provider/router.mjs +242 -241
  74. package/src/provider/sse.mjs +104 -104
  75. package/src/provider/wizard.mjs +556 -0
  76. package/src/repl/capability-facade.mjs +30 -0
  77. package/src/repl/command-surface.mjs +23 -0
  78. package/src/repl/controller-entry.mjs +40 -0
  79. package/src/repl/core-shell.mjs +208 -0
  80. package/src/repl/dialog-router.mjs +87 -0
  81. package/src/repl/input-engine.mjs +76 -0
  82. package/src/repl/keymap.mjs +7 -0
  83. package/src/repl/operator-surface.mjs +15 -0
  84. package/src/repl/permission-flow.mjs +49 -0
  85. package/src/repl/runtime-facade.mjs +36 -0
  86. package/src/repl/slash-router.mjs +62 -0
  87. package/src/repl/state-store.mjs +29 -0
  88. package/src/repl/turn-controller.mjs +58 -0
  89. package/src/repl/verification.mjs +23 -0
  90. package/src/repl.mjs +3368 -2929
  91. package/src/rules/load-rules.mjs +3 -3
  92. package/src/runtime.mjs +1 -1
  93. package/src/session/agent-transaction.mjs +86 -0
  94. package/src/session/checkpoint.mjs +302 -302
  95. package/src/session/compaction.mjs +36 -14
  96. package/src/session/engine.mjs +417 -227
  97. package/src/session/longagent-4stage.mjs +467 -460
  98. package/src/session/longagent-hybrid.mjs +1344 -1081
  99. package/src/session/longagent-plan.mjs +376 -365
  100. package/src/session/longagent-project-memory.mjs +53 -53
  101. package/src/session/longagent-scaffold.mjs +291 -291
  102. package/src/session/longagent-task-bus.mjs +138 -54
  103. package/src/session/longagent-utils.mjs +828 -472
  104. package/src/session/longagent.mjs +911 -884
  105. package/src/session/loop.mjs +1005 -905
  106. package/src/session/prompt/agent.txt +25 -0
  107. package/src/session/prompt/anthropic.txt +150 -150
  108. package/src/session/prompt/beast.txt +1 -1
  109. package/src/session/prompt/plan.txt +28 -6
  110. package/src/session/prompt/qwen.txt +46 -46
  111. package/src/session/recovery.mjs +21 -0
  112. package/src/session/rollback.mjs +197 -0
  113. package/src/session/routing-observability.mjs +72 -0
  114. package/src/session/runtime-state.mjs +47 -0
  115. package/src/session/store.mjs +523 -510
  116. package/src/session/system-prompt.mjs +56 -8
  117. package/src/session/task-validator.mjs +267 -267
  118. package/src/session/usability-gates.mjs +2 -2
  119. package/src/skill/builtin/commit.mjs +64 -64
  120. package/src/skill/builtin/design.mjs +76 -76
  121. package/src/skill/generator.mjs +18 -2
  122. package/src/skill/registry.mjs +642 -390
  123. package/src/storage/audit-store.mjs +18 -11
  124. package/src/storage/event-log.mjs +7 -1
  125. package/src/storage/ghost-commit-store.mjs +243 -245
  126. package/src/storage/paths.mjs +13 -0
  127. package/src/theme/default-theme.mjs +1 -1
  128. package/src/theme/markdown.mjs +4 -0
  129. package/src/theme/schema.mjs +1 -1
  130. package/src/theme/status-bar.mjs +162 -158
  131. package/src/tool/audit-wrapper.mjs +18 -2
  132. package/src/tool/edit-transaction.mjs +23 -0
  133. package/src/tool/executor.mjs +26 -1
  134. package/src/tool/file-read-state.mjs +65 -0
  135. package/src/tool/git-auto.mjs +526 -526
  136. package/src/tool/git-full-auto.mjs +487 -478
  137. package/src/tool/mutation-guard.mjs +54 -0
  138. package/src/tool/prompt/edit.txt +3 -3
  139. package/src/tool/prompt/multiedit.txt +1 -0
  140. package/src/tool/prompt/notebookedit.txt +2 -1
  141. package/src/tool/prompt/patch.txt +25 -24
  142. package/src/tool/prompt/read.txt +3 -3
  143. package/src/tool/prompt/sysinfo.txt +29 -0
  144. package/src/tool/prompt/task.txt +66 -4
  145. package/src/tool/prompt/write.txt +2 -2
  146. package/src/tool/question-prompt.mjs +17 -4
  147. package/src/tool/registry.mjs +1701 -1343
  148. package/src/tool/task-tool.mjs +14 -6
  149. package/src/ui/activity-renderer.mjs +667 -664
  150. package/src/ui/repl-background-panel.mjs +7 -0
  151. package/src/ui/repl-capability-panel.mjs +9 -0
  152. package/src/ui/repl-dashboard.mjs +54 -4
  153. package/src/ui/repl-help.mjs +110 -0
  154. package/src/ui/repl-operator-panel.mjs +12 -0
  155. package/src/ui/repl-route-feedback.mjs +35 -0
  156. package/src/ui/repl-status-view.mjs +76 -0
  157. package/src/ui/repl-task-panel.mjs +5 -0
  158. package/src/ui/repl-transcript-panel.mjs +56 -0
  159. package/src/ui/repl-turn-summary.mjs +135 -0
  160. package/src/usage/pricing.mjs +122 -121
  161. package/src/usage/usage-meter.mjs +1 -0
  162. package/src/util/git.mjs +562 -519
  163. package/src/util/template.mjs +6 -1
package/README.md CHANGED
@@ -1,387 +1,452 @@
1
- # kkcode
2
-
3
- [![npm version](https://img.shields.io/npm/v/@kkelly-offical/kkcode)](https://www.npmjs.com/package/@kkelly-offical/kkcode)
4
- ![Node](https://img.shields.io/badge/Node.js-%3E%3D22-green)
5
- ![License](https://img.shields.io/badge/License-GPL--3.0-blue)
6
- ![Platform](https://img.shields.io/badge/Platform-Windows%20%7C%20macOS%20%7C%20Linux-lightgrey)
7
- ![Models](https://img.shields.io/badge/Models-Qwen%20%7C%20GLM%20%7C%20DeepSeek%20%7C%20Claude%20%7C%20GPT%20%7C%20Ollama-orange)
8
-
9
- 面向团队协作的终端 AI Coding Agent
10
-
11
- **可执行** 内置 16+ 工具链,子智能体委派,主动规划与审批。
12
- **可治理** 三级权限策略,会话审计,Token 预算控制。
13
- **可长跑** — LongAgent 阶段并行编排,自动重试与门禁闭环。
14
- **多模型** — 原生支持 Qwen3.5、GLM-5、DeepSeek、Claude、GPT、Ollama 本地模型。
15
-
16
- ---
17
-
18
- ## 目录
19
-
20
- - [快速开始](#快速开始)
21
- - [支持的模型](#支持的模型)
22
- - [模式系统](#模式系统)
23
- - [LongAgent 编排](#longagent-编排)
24
- - [主动规划](#主动规划)
25
- - [权限系统](#权限系统)
26
- - [内置工具](#内置工具)
27
- - [子智能体](#子智能体)
28
- - [Auto Memory](#auto-memory)
29
- - [MCP 接入](#mcp-接入)
30
- - [GitHub 集成](#github-集成)
31
- - [扩展机制](#扩展机制)
32
- - [会话与审计](#会话与审计)
33
- - [TUI 交互](#tui-交互)
34
- - [项目结构](#项目结构)
35
- - [配置](#配置)
36
- - [常见问题](#常见问题)
37
- - [致谢](#致谢)
38
- - [License](#license)
39
-
40
- ---
41
-
42
- ## 快速开始
43
-
44
- ### 环境要求
45
-
46
- - Node.js `>=22`
47
- - npm / pnpm
48
- - 建议终端:Windows Terminal、iTerm2 或现代 Linux terminal
49
-
50
- ### 安装与运行
51
-
52
- **从 npm 安装(推荐):**
53
-
54
- ```bash
55
- npm install -g @kkelly-offical/kkcode
56
- kkcode
57
- ```
58
-
59
- **从源码运行(开发用):**
60
-
61
- ```bash
62
- git clone https://github.com/kkelly-offical/kkcode.git
63
- cd kkcode
64
- npm install
65
- npm run start
66
- ```
67
-
68
- ### 初始化项目配置
69
-
70
- ```bash
71
- kkcode init -y
72
- ```
73
-
74
- 配置文件按优先级自动查找:
75
-
76
- - 用户级:`~/.kkcode/config.yaml`
77
- - 项目级:`./kkcode.config.yaml` `./.kkcode/config.yaml`
78
-
79
- ---
80
-
81
- ## 支持的模型
82
-
83
- kkcode 通过统一的 Provider 抽象层支持多种模型,开箱即用:
84
-
85
- | 提供商 | 模型 | 类型 | 配置样例 |
86
- |--------|------|------|----------|
87
- | 阿里 DashScope | Qwen3.5-Plus、Qwen3-Max、Qwen3-Coder | `openai-compatible` | [`config-qwen3.5.yaml`](configs/config-qwen3.5.yaml) |
88
- | 智谱 AI | GLM-5、GLM-4.5 | `openai-compatible` | [`config-glm5.yaml`](configs/config-glm5.yaml) |
89
- | DeepSeek | DeepSeek-Chat (V3)、DeepSeek-Reasoner (R1) | `openai-compatible` | [`config-deepseek.yaml`](configs/config-deepseek.yaml) |
90
- | Anthropic | Claude Opus 4.6、Sonnet 4.6 | `anthropic` | [`config.example.yaml`](docs/config.example.yaml) |
91
- | OpenAI | GPT-5.3-Codex、GPT-5.2 | `openai` | [`config.example.yaml`](docs/config.example.yaml) |
92
- | Ollama | Qwen3、DeepSeek-Coder、LLaMA 等本地模型 | `ollama` | [`config-ollama.yaml`](configs/config-ollama.yaml) |
93
- | 任意 OpenAI 兼容 API | Kimi、Yi、Moonshot 等 | `openai-compatible` | [`config-multi-provider.yaml`](configs/config-multi-provider.yaml) |
94
-
95
- 切换模型只需修改配置文件中的 `provider.default`,或在 TUI 中使用 `/provider` 和 `/model` 命令。
96
-
97
- ---
98
-
99
- ## 模式系统
100
-
101
- | 模式 | 目标 | 工具权限 | 典型场景 |
102
- |---|---|---|---|
103
- | `ask` | 问答 / 解释 | 只读 | 理解代码、解释报错 |
104
- | `plan` | 方案拆解 | 只读 | 先出执行计划 |
105
- | `agent` | 单轮执行 | 全工具 | 快速改代码 + 运行命令 |
106
- | `longagent` | 长程编排 | 全工具 + 调度 | 跨文件 / 多阶段任务 |
107
-
108
- TUI 中按 `Tab` 一键轮换模式。
109
-
110
- ---
111
-
112
- ## LongAgent 编排
113
-
114
- LongAgent kkcode 的核心差异能力,支持两种编排模式:
115
-
116
- **Hybrid 模式**(默认)— 7 阶段流水线:
117
-
118
- ```
119
- Intake → Preview → Blueprint → Git分支 → Scaffold+并行编码 → Debugging → 门禁验证 → Git合并
120
- ```
121
-
122
- **4-Stage 模式** 4 阶段顺序执行:
123
-
124
- ```
125
- Preview(只读) Blueprint(只读) Coding(写入) → Debugging(写入)
126
- ```
127
-
128
- ### 关键机制
129
-
130
- - **阶段并行** — 同阶段任务由独立 worker 并发执行,`max_concurrency` 控制并发数
131
- - **文件隔离** — 每个文件仅归属一个 task,plan 阶段即检测冲突
132
- - **自动重试** 失败任务自动重试,优先续写未完成文件
133
- - **门禁闭环** — build / test / review / health / budget 五项质量门禁
134
- - **Git 集成** 自动创建特性分支,每阶段提交,完成后合并
135
- - **防卡死** 检测探索循环,强制推进到下一阶段
136
-
137
- ---
138
-
139
- ## 主动规划
140
-
141
- Agent 执行中可主动进入规划模式,无需用户手动切换:
142
-
143
- 1. Agent 遇到复杂任务时调用 `enter_plan`
144
- 2. 在规划模式下分析代码、设计方案
145
- 3. 调用 `exit_plan` 提交计划,TUI 弹出审批面板
146
- 4. 用户 Approve 或 Reject(可附反馈)
147
- 5. 批准后按计划执行,驳回则修订
148
-
149
- ---
150
-
151
- ## 权限系统
152
-
153
- 采用 **策略 + 交互审批 + 会话缓存授权** 组合模型。
154
-
155
- - `permission.default_policy`: `ask | allow | deny`
156
- - `permission.rules[]`: 按工具 / 模式 / 文件模式 / 命令前缀匹配细粒度规则
157
- - 审批一次后同会话内同类操作自动放行
158
-
159
- 提供三级权限模板:[严格](configs/permission-strict.yaml) | [标准](configs/permission-standard.yaml) | [宽松](configs/permission-permissive.yaml)
160
-
161
- TUI 审批面板:`1` allow once / `2` allow session / `3` deny / `Esc` deny
162
-
163
- ---
164
-
165
- ## 内置工具
166
-
167
- | 工具 | 说明 |
168
- |------|------|
169
- | `read` | 读取文件内容(支持 offset/limit 分页) |
170
- | `write` | 原子写文件(overwrite / append / insert) |
171
- | `edit` | 事务性字符串替换 + 自动回滚 |
172
- | `patch` | 按行号范围替换文件内容 |
173
- | `glob` | 按模式搜索文件 |
174
- | `grep` | 按正则搜索文件内容 |
175
- | `bash` | 执行 shell 命令 |
176
- | `task` | 委派子智能体执行子任务 |
177
- | `todowrite` | 结构化任务管理 |
178
- | `question` | 向用户提问 |
179
- | `enter_plan` / `exit_plan` | 主动规划与审批 |
180
- | `webfetch` / `websearch` | 网页抓取与搜索 |
181
- | `codesearch` | 代码搜索引擎 |
182
-
183
- 写入安全特性:原子写、事务回滚、外部修改检测、读前编辑约束、file lock 串行化。
184
-
185
- ---
186
-
187
- ## 子智能体
188
-
189
- 通过 `task` 工具委派专项子智能体:
190
-
191
- | 类型 | 说明 | 权限 |
192
- |------|------|------|
193
- | `build` | 通用构建执行 | 全工具 |
194
- | `explore` | 快速代码探索 | 只读 |
195
- | `reviewer` | 代码审查(bug / 安全 / 质量) | 只读 |
196
- | `researcher` | 深度研究 + Web 搜索 | 只读 + 网络 |
197
-
198
- 支持通过 YAML/MJS 自定义子智能体,或 `/create-agent` 自动生成。
199
-
200
- ---
201
-
202
- ## Auto Memory
203
-
204
- 为每个项目维护独立持久记忆,跨会话保存项目知识。
205
-
206
- - 存储位置:`~/.kkcode/projects/<项目名>_<hash>/memory/MEMORY.md`
207
- - 每次会话启动时自动注入系统提示词(限 200 行)
208
- - Agent 可直接读写记忆文件
209
- - 状态栏显示 `MEM` 徽章表示已加载
210
-
211
- ---
212
-
213
- ## MCP 接入
214
-
215
- 支持三种传输协议:
216
-
217
- | 传输 | 适用场景 | 关键字段 |
218
- |------|----------|----------|
219
- | `stdio` | 本地子进程 | `command`, `args` |
220
- | `sse` | 远程 Streamable HTTP | `url`, `headers` |
221
- | `http` | REST 风格 | `url`, `headers` |
222
-
223
- 自动发现:合并 `.mcp.json`、`.kkcode/mcp.json`、`~/.kkcode/mcp.json` 中的配置。
224
-
225
- ```bash
226
- kkcode mcp test # 健康状态
227
- kkcode mcp tools # 可用工具列表
228
- ```
229
-
230
- ---
231
-
232
- ## GitHub 集成
233
-
234
- 通过 GitHub Device Flow 实现安全的仓库访问,无需手动配置 Token。
235
-
236
- ```bash
237
- kkcode --github # 登录并选择仓库
238
- kkcode --github logout # 登出
239
- ```
240
-
241
- 工作流程:登录 → 选择仓库/分支 → 选择本地或云端模式 → 进入 REPL → 退出时询问是否推送。
242
-
243
- ---
244
-
245
- ## 扩展机制
246
-
247
- | 扩展类型 | 目录 | 说明 |
248
- |----------|------|------|
249
- | 命令模板 | `.kkcode/commands/` | Markdown 模板,支持变量替换 |
250
- | 技能 | `.kkcode/skills/` | 可编程技能,`/create-skill` 生成 |
251
- | 子智能体 | `.kkcode/agents/` | YAML/MJS 定义,`/create-agent` 生成 |
252
- | 自定义工具 | `.kkcode/tools/` | .mjs 自动加载,`/reload` 热更新 |
253
- | 插件/Hook | `.kkcode/plugins/` | Hook 事件脚本 |
254
- | 规则 | `.kkcode/rules/` | 项目级提示词规则 |
255
- | 指令文件 | `KKCODE.md` | 项目级指令,自动注入提示词 |
256
-
257
- ---
258
-
259
- ## 会话与审计
260
-
261
- - 会话存储:`~/.kkcode/sessions/`
262
- - 审计日志:`~/.kkcode/audit-log.json`
263
- - 事件日志:`~/.kkcode/events.log`
264
- - 三层用量追踪:turn / session / global
265
- - 预算策略:`warn | block`
266
-
267
- ```bash
268
- kkcode doctor --json # 完整诊断
269
- kkcode session list # 会话列表
270
- kkcode session gc # 过期清理
271
- kkcode usage show # 用量统计
272
- kkcode longagent status # LongAgent 状态
273
- ```
274
-
275
- ---
276
-
277
- ## TUI 交互
278
-
279
- ### 状态栏
280
-
281
- 底部实时显示:`MODE` `MODEL` `TOKENS` `COST` `CONTEXT` `MEM` `PERMISSION` `LONG`
282
-
283
- ### 快捷键
284
-
285
- | 按键 | 功能 |
286
- |------|------|
287
- | `Enter` | 发送 |
288
- | `Ctrl+J` | 换行 |
289
- | `Tab` | 模式轮换 |
290
- | `Up/Down` | 输入历史 |
291
- | `Ctrl+Up/Down` | 滚动日志区 |
292
- | `Esc` | 清空输入 |
293
- | `Ctrl+L` | 清空活动区 |
294
- | `Ctrl+C` | 退出 |
295
-
296
- 支持 `@图片路径` `@图片URL` 引用多模态输入。
297
-
298
- ### 常用 Slash 命令
299
-
300
- `/help` `/status` `/history` `/new` `/resume` `/mode` `/provider` `/model` `/permission` `/paste` `/commands` `/reload` `/clear` `/exit`
301
-
302
- ---
303
-
304
- ## 项目结构
305
-
306
- ```
307
- kkcode/
308
- ├── src/
309
- │ ├── index.mjs # CLI 入口
310
- │ ├── repl.mjs # TUI 主循环
311
- │ ├── core/ # 核心类型、常量、事件总线
312
- │ ├── config/ # 配置加载、Schema 校验、默认值
313
- │ ├── session/ # 会话引擎、消息循环、LongAgent 编排
314
- │ ├── tool/ # 工具注册、执行、事务管理
315
- │ ├── agent/ # 子智能体(explore/reviewer/researcher)
316
- │ ├── provider/ # 多 Provider 适配
317
- │ ├── permission/ # 权限引擎
318
- │ ├── mcp/ # MCP 客户端(stdio/sse/http)
319
- │ ├── orchestration/ # 后台任务管理与并行 worker
320
- │ ├── observability/ # 可观测性(Metrics/Tracer)
321
- │ ├── ui/ # Dashboard 渲染
322
- │ ├── storage/ # 会话/任务持久化
323
- │ ├── usage/ # Token 计量与预算
324
- │ └── util/ # 通用工具函数
325
- ├── configs/ # 配置样例与权限模板
326
- ├── test/ # 单元测试与集成测试
327
- ├── package.json
328
- ├── LICENSE # GPL-3.0
329
- └── NOTICE.md # 第三方致谢与声明
330
- ```
331
-
332
- ---
333
-
334
- ## 配置
335
-
336
- 完整配置参考:[`docs/config.example.yaml`](docs/config.example.yaml)
337
-
338
- ### 配置样例(configs/ 目录)
339
-
340
- | 文件 | 说明 |
341
- |------|------|
342
- | [`config-qwen3.5.yaml`](configs/config-qwen3.5.yaml) | 通义千问 Qwen3.5 系列 |
343
- | [`config-qwen3.yaml`](configs/config-qwen3.yaml) | 通义千问 Qwen3 系列 |
344
- | [`config-glm5.yaml`](configs/config-glm5.yaml) | 智谱 GLM-5 / GLM-4.5 |
345
- | [`config-deepseek.yaml`](configs/config-deepseek.yaml) | DeepSeek V3 / R1 |
346
- | [`config-ollama.yaml`](configs/config-ollama.yaml) | 本地 Ollama |
347
- | [`config-multi-provider.yaml`](configs/config-multi-provider.yaml) | 多 Provider 组合 |
348
- | [`permission-strict.yaml`](configs/permission-strict.yaml) | 严格权限(生产环境) |
349
- | [`permission-standard.yaml`](configs/permission-standard.yaml) | 标准权限(日常开发) |
350
- | [`permission-permissive.yaml`](configs/permission-permissive.yaml) | 宽松权限(个人/CI) |
351
-
352
- ---
353
-
354
- ## 常见问题
355
-
356
- **Q: LongAgent 为什么拒绝"你好"?**
357
- LongAgent 是执行型编排器,非编码目标会被意图识别拦截。请使用 `ask` 模式进行对话。
358
-
359
- **Q: enter_plan /plan 有什么区别?**
360
- `/plan` 是用户手动切换模式。`enter_plan` 是 agent 执行中主动进入规划,审批通过后继续执行。
361
-
362
- **Q: 如何切换模型?**
363
- TUI 中使用 `/provider` 切换提供商,`/model` 切换模型。或修改配置文件中的 `provider.default`。
364
-
365
- **Q: Auto Memory 会无限增长吗?**
366
- 注入系统提示时限制 200 行。Agent 会保持精简并删除过时条目。
367
-
368
- **Q: 支持哪些国产模型?**
369
- 原生支持通义千问 Qwen3.5/3 系列、智谱 GLM-5/4.5、DeepSeek V3/R1,以及任何 OpenAI 兼容 API。
370
-
371
- ---
372
-
373
- ## 致谢
374
-
375
- kkcode 的设计受到以下项目的启发:
376
-
377
- - **[Claude Code](https://github.com/anthropics/claude-code)** — Anthropic 官方 AI Coding CLI。工具体系、子智能体架构、提示词工程等核心设计以此为标杆。
378
- - **[OpenCode](https://github.com/nicepkg/opencode)** 开源终端 AI Coding 助手。多 Provider 支持、主题系统等借鉴了其设计。
379
- - **[Everything Claude Code](https://github.com/affaan-m/everything-claude-code)** — Instinct 自动学习、Hook Recipes、TDD 工作流等能力受此启发。
380
-
381
- ---
382
-
383
- ## License
384
-
385
- 本项目基于 [GNU General Public License v3.0](LICENSE) 开源。
386
-
387
- Copyright (C) 2026 kkcode team
1
+ # kkcode
2
+
3
+ [![npm version](https://img.shields.io/npm/v/@kkelly-offical/kkcode?label=v0.2.1)](https://www.npmjs.com/package/@kkelly-offical/kkcode)
4
+ [![GitHub Release](https://img.shields.io/github/v/release/kkelly-offical/kkcode)](https://github.com/kkelly-offical/kkcode/releases)
5
+ ![Node](https://img.shields.io/badge/Node.js-%3E%3D22-green)
6
+ ![License](https://img.shields.io/badge/License-GPL--3.0-blue)
7
+ ![Platform](https://img.shields.io/badge/Platform-Windows%20%7C%20macOS%20%7C%20Linux-lightgrey)
8
+
9
+ **Terminal-first personal assistant with dedicated Coding Agent and LongAgent modes for governed execution and extensible local workflows.**
10
+
11
+ **终端优先、可治理、可扩展的个人助手,内置专门 Coding Agent LongAgent 模式。**
12
+ kkcode 把问答、规划、事务型修改、多阶段长任务编排放在同一个 CLI 工作台里,并且把权限、预算、审计、后台任务、MCP、技能与插件一起纳入统一执行面。
13
+
14
+ ---
15
+
16
+ <a id="table-of-contents"></a>
17
+ ## Table of Contents / 目录
18
+
19
+ - [Overview / 概览](#overview)
20
+ - [Why kkcode / 为什么选择 kkcode](#why-kkcode)
21
+ - [Installation / 安装](#installation)
22
+ - [Quick Start / 快速开始](#quick-start)
23
+ - [Capability Snapshot / 能力总览](#capability-snapshot)
24
+ - [Modes & LongAgent / 模式与 LongAgent](#modes-and-longagent)
25
+ - [Safety & Permissions / 权限与安全](#safety-and-permissions)
26
+ - [Delegation & Subagents / 委派与子智能体](#delegation-and-subagents)
27
+ - [Integrations / 集成](#integrations)
28
+ - [Extensions / 扩展机制](#extensions)
29
+ - [TUI & CLI Reference / TUI 与命令参考](#tui-and-cli-reference)
30
+ - [Configuration & Project Layout / 配置与项目结构](#configuration-and-project-layout)
31
+ - [Release Status / 发布状态](#release-status)
32
+ - [Compatibility, Limits & Roadmap / 兼容性、边界与路线图](#compatibility-limits-and-roadmap)
33
+ - [FAQ / 常见问题](#faq)
34
+ - [Contributing / 贡献](#contributing)
35
+ - [License / 许可证](#license)
36
+ - [Further Reading / 延伸阅读](#further-reading)
37
+
38
+ ---
39
+
40
+ <a id="overview"></a>
41
+ ## Overview / 概览
42
+
43
+ **English**
44
+ - kkcode is a terminal-native personal assistant designed for local work, governed execution, coding, and multi-stage delivery.
45
+ - It keeps four public lanes — `assistant`, `plan`, `agent`/`code`, and `longagent` — under one CLI surface.
46
+ - It is optimized for **CLI-first** and **LongAgent-first** workflows rather than GUI-first or marketplace-first product patterns.
47
+
48
+ **中文**
49
+ - kkcode 是一个面向终端原生工作流的个人助手,强调本地事务、可治理执行、编码和多阶段交付。
50
+ - 它把四条公开执行航道 —— `assistant`、`plan`、`agent`/`code`、`longagent` —— 收敛在同一个 CLI 入口下。
51
+ - 它优先服务 **CLI-first**、**LongAgent-first** 的工程工作流,而不是 GUI 优先或 marketplace 优先的平台形态。
52
+
53
+ ---
54
+
55
+ <a id="why-kkcode"></a>
56
+ ## Why kkcode / 为什么选择 kkcode
57
+
58
+ **English**
59
+ - **CLI-first**: core workflows stay in the terminal.
60
+ - **LongAgent-first**: large tasks are planned, staged, and verified instead of improvised in one prompt.
61
+ - **Governed execution**: permissions, budgets, audit logs, and recovery are built in.
62
+ - **Local extensibility**: MCP, skills, commands, hooks, tools, and custom agents can all be mounted locally.
63
+
64
+ **中文**
65
+ - **CLI-first**:核心工作流都在终端内完成。
66
+ - **LongAgent-first**:复杂任务先规划、分阶段、带门禁,而不是靠单轮 prompt 硬顶。
67
+ - **可治理执行**:权限、预算、审计、恢复、后台任务都是内建能力。
68
+ - **本地可扩展**:MCP、skills、commands、hooks、tools、custom agents 都能本地挂载。
69
+
70
+ ---
71
+
72
+ <a id="installation"></a>
73
+ ## Installation / 安装
74
+
75
+ **Requirements / 环境要求**
76
+ - Node.js `>=22`
77
+ - npm or pnpm
78
+ - A modern terminal on Windows, macOS, or Linux
79
+
80
+ **Install from npm / 通过 npm 安装**
81
+ ```bash
82
+ npm install -g @kkelly-offical/kkcode
83
+ kkcode
84
+ ```
85
+
86
+ **Run from source / 从源码运行**
87
+ ```bash
88
+ git clone https://github.com/kkelly-offical/kkcode.git
89
+ cd kkcode
90
+ npm install
91
+ npm run start
92
+ ```
93
+
94
+ **Useful links / 常用链接**
95
+ - [npm package](https://www.npmjs.com/package/@kkelly-offical/kkcode)
96
+ - [GitHub Releases](https://github.com/kkelly-offical/kkcode/releases)
97
+ - [Example config](docs/config.example.yaml)
98
+
99
+ ---
100
+
101
+ <a id="quick-start"></a>
102
+ ## Quick Start / 快速开始
103
+
104
+ **1. Launch / 启动**
105
+ ```bash
106
+ kkcode
107
+ ```
108
+
109
+ **2. Initialize project config / 初始化项目配置**
110
+ ```bash
111
+ kkcode init -y
112
+ ```
113
+
114
+ **3. Verify the install / 验证安装**
115
+ ```bash
116
+ kkcode --help
117
+ kkcode doctor
118
+ ```
119
+
120
+ **First-run behavior / 首次启动行为**
121
+ - On first launch, kkcode runs onboarding and records your preferences.
122
+ - Use `/profile` to inspect or update personal preferences.
123
+ - Use `/like` to rerun onboarding.
124
+
125
+ **Configuration search order / 配置查找顺序**
126
+ - User-level: `~/.kkcode/config.yaml`
127
+ - Project-level: `./kkcode.config.yaml` or `./.kkcode/config.yaml`
128
+
129
+ ---
130
+
131
+ <a id="capability-snapshot"></a>
132
+ ## Capability Snapshot / 能力总览
133
+
134
+ | Area / 能力面 | Status / 状态 | Notes / 说明 |
135
+ | --- | --- | --- |
136
+ | Assistant / 个人助手 | Supported | Default CLI lane for terminal-native personal work, explanation, and code understanding |
137
+ | Plan / 方案规划 | Supported | Planning without mutating the repo |
138
+ | Agent / 默认事务航道 | Supported | Local inspect/patch/verify loops |
139
+ | LongAgent / 长程编排 | Supported | Multi-stage execution, retries, gates, resumable flow |
140
+ | Permissions / 权限治理 | Supported | Policy + approvals + session cache |
141
+ | Background tasks / 后台任务 | Supported | Launch, inspect, wait, retry, cancel |
142
+ | MCP / 模型上下文协议 | Supported | Local MCP discovery and registry |
143
+ | Skills / Commands / Hooks | Supported | Local-first extensibility surface |
144
+ | Plugins / 插件包 | MVP | Local plugin manifests and component toggles |
145
+ | GUI / IDE / desktop automation | Not promised | README does not claim GUI-first product support |
146
+
147
+ For a deeper boundary matrix, see [CLI General Assistant Capability Matrix](docs/cli-general-assistant-capability-matrix.md).
148
+
149
+ ---
150
+
151
+ <a id="modes-and-longagent"></a>
152
+ ## Modes & LongAgent / 模式与 LongAgent
153
+
154
+ ### Public lanes / 公开执行航道
155
+
156
+ | Mode | Purpose | Typical use |
157
+ | --- | --- | --- |
158
+ | `assistant` | default personal assistant | local files, logs, system checks, web lookup, Git/GitHub, notes, tasks |
159
+ | `plan` | specification / planning | producing an execution plan before mutations |
160
+ | `agent` / `code` / `coding` | dedicated coding execution | inspect + patch + verify small/medium coding tasks |
161
+ | `longagent` | staged orchestration | multi-file, multi-step, ownership-driven delivery |
162
+
163
+ **English**
164
+ - `assistant` is the default terminal personal-assistant lane.
165
+ - `agent` / `code` / `coding` is the dedicated coding lane.
166
+ - Only escalate to `longagent` when the task is clearly multi-stage or system-wide.
167
+ - Interrupted work can be resumed with the same session context.
168
+
169
+ **中文**
170
+ - `assistant` 是默认的终端个人助手航道。
171
+ - `agent` / `code` / `coding` 是专门编码航道。
172
+ - 只有在任务明显跨文件、跨阶段、影响面较大时,才建议升级到 `longagent`。
173
+ - 中断后的工作可以在同一会话中继续,不需要从零开始。
174
+ - **路由理由可见**:当 kkcode 自动建议模式变化时,会尽量解释为什么当前任务更适合留在 `assistant` / `agent` 或升级到 `longagent`。
175
+
176
+ ### CLI 通用助手能力边界(0.1.13)
177
+
178
+ **公共模式契约**
179
+
180
+ - `assistant`:默认终端个人助手航道,承接本地文件、日志、系统信息、网页查询、Git/GitHub、笔记、任务整理、解释、答疑和分析。
181
+ - `plan`:**只产出规格,不执行文件变更**。
182
+ - `agent` / `code` / `coding`:专门编码航道,承接 inspect / patch / verify 小闭环事务。
183
+ - **只有出现明确重型证据时,才从 `assistant` 或 `agent` 升级到 `longagent`**。
184
+
185
+ **能力边界速览**
186
+ - 系统 / 运行时信息
187
+ - 本地目录 / 文件 / 日志检查
188
+ - 仓库 / 发布辅助
189
+ - 这**不代表** kkcode 已经承诺 GUI / 桌面自动化能力
190
+ - 默认先在 `assistant` 内处理普通终端事务;明确编码修改进入 `agent` / `code`,再判断是否需要升级
191
+
192
+ **Further reading / 延伸阅读**
193
+ - [0.1.13 Mode Lane Contract](docs/kkcode-0.1.13-mode-lane-contract.md)
194
+ - [Agent Mode Tolerance Contract](docs/kkcode-0.1.12-agent-mode-tolerance-contract.md)
195
+
196
+ ---
197
+
198
+ <a id="safety-and-permissions"></a>
199
+ ## Safety & Permissions / 权限与安全
200
+
201
+ **English**
202
+ - kkcode uses a policy-driven permission model with optional approvals.
203
+ - Session-scoped grants can reduce repeated prompts while preserving boundaries.
204
+ - Budget and usage controls are designed to keep long-running sessions governable.
205
+
206
+ **中文**
207
+ - kkcode 使用策略驱动的权限模型,并可叠加交互式审批。
208
+ - 会话级授权缓存可减少重复确认,同时保持边界清晰。
209
+ - 预算与用量控制让长会话、长任务仍然处于可治理状态。
210
+
211
+ **Policy examples / 策略示例**
212
+ - `permission.default_policy: ask | allow | deny`
213
+ - rule-based overrides by tool / mode / file pattern / command prefix
214
+
215
+ ---
216
+
217
+ <a id="delegation-and-subagents"></a>
218
+ ## Delegation & Subagents / 委派与子智能体
219
+
220
+ **English**
221
+ - kkcode supports bounded delegation through the `task` surface.
222
+ - Use `fresh_agent` for isolated implementation work.
223
+ - Use `fork_context` for read-only sidecar work such as research or verification.
224
+ - Do not outsource core understanding when the main thread must synthesize the result.
225
+
226
+ **中文**
227
+ - kkcode 通过 `task` 能力支持有边界的委派。
228
+ - `fresh_agent` 适合隔离实现任务。
229
+ - `fork_context` 适合研究、审计、验证这类只读 sidecar 任务。
230
+ - 如果主线程必须综合判断,就不要把理解工作本身外包出去。
231
+
232
+ **Background task contract / 后台任务契约**
233
+ - 通过 `background_output` 查看后台任务输出
234
+ - 通过 `background_cancel` 取消后台任务
235
+ - 终态固定为 `completed` / `cancelled` / `error` / `interrupted`
236
+
237
+ **Further reading / 延伸阅读**
238
+ - [Task Delegation Contract Matrix](docs/task-delegation-contract-matrix.md)
239
+ - [Agent / LongAgent Extension Guide](docs/agent-longagent-compat-extension-guide.md)
240
+
241
+ ---
242
+
243
+ <a id="integrations"></a>
244
+ ## Integrations / 集成
245
+
246
+ ### MCP
247
+ - Discover local MCP definitions and mount tools into the runtime.
248
+ - Inspect registered MCP servers from the CLI.
249
+ - Use MCP as part of the same governed tool surface.
250
+
251
+ ### GitHub
252
+ - Authenticate, inspect repositories, and run GitHub-related flows from the terminal.
253
+ - Repository helpers live under `src/github/`.
254
+
255
+ ### Git automation
256
+ - Local git-aware helpers support safe status, patch, and snapshot workflows.
257
+ - See [GIT_AUTO_USAGE.md](docs/GIT_AUTO_USAGE.md).
258
+
259
+ ---
260
+
261
+ <a id="extensions"></a>
262
+ ## Extensions / 扩展机制
263
+
264
+ **Local-first extension surface / 本地优先扩展面**
265
+ - commands
266
+ - skills
267
+ - agents
268
+ - tools
269
+ - hooks
270
+ - plugin manifests
271
+
272
+ **Directory conventions / 目录约定**
273
+ - `.kkcode/commands/`
274
+ - `.kkcode/skills/`
275
+ - `.kkcode/agents/`
276
+ - `.kkcode/tools/`
277
+ - `.kkcode/plugins/`
278
+ - `.kkcode/hooks/`
279
+ - `.kkcode-plugin/plugin.json`
280
+
281
+ **English**
282
+ - kkcode’s extension story is local-first and explicit.
283
+ - Plugins are currently an MVP surface, not a marketplace platform promise.
284
+
285
+ **中文**
286
+ - kkcode 的扩展机制是本地优先、显式可控的。
287
+ - 当前插件能力是 MVP,不代表已经承诺 marketplace 平台形态。
288
+
289
+ **Further reading / 延伸阅读**
290
+ - [ClaudeNext Agent / LongAgent Skills Compatibility](docs/claudenext-agent-longagent-skills-compat.md)
291
+ - [Agent / LongAgent Extension Guide](docs/agent-longagent-compat-extension-guide.md)
292
+
293
+ ---
294
+
295
+ <a id="tui-and-cli-reference"></a>
296
+ ## TUI & CLI Reference / TUI 与命令参考
297
+
298
+ ### Common TUI slash commands / 常用 TUI slash 命令
299
+ - `/help` — show help
300
+ - `/status` show runtime and operator status
301
+ - `/commands` — inspect command / skill / capability surface
302
+ - `/reload` — reload commands, skills, and agents
303
+ - `/new`, `/resume`, `/history` — session lifecycle
304
+ - `/provider`, `/model` — provider/model switching
305
+ - `/permission` — permission policy management
306
+ - `/create-skill`, `/create-agent` — generate local extensions
307
+
308
+ **Interrupt semantics / 中断语义**
309
+ - `Esc` 可用于**中断当前 turn**、退出部分选择态或拒绝当前交互式请求,具体行为取决于当前上下文。
310
+
311
+ ### Main CLI commands / 主要 CLI 子命令
312
+ - `chat`
313
+ - `session`
314
+ - `background`
315
+ - `agent`
316
+ - `longagent`
317
+ - `mcp`
318
+ - `skill`
319
+ - `config`
320
+ - `doctor`
321
+ - `usage`
322
+ - `review`
323
+ - `audit`
324
+
325
+ Run `kkcode --help` or `kkcode <command> --help` for the full surface.
326
+
327
+ ---
328
+
329
+ <a id="configuration-and-project-layout"></a>
330
+ ## Configuration & Project Layout / 配置与项目结构
331
+
332
+ ### Key config themes / 关键配置主题
333
+ - provider/model selection
334
+ - permission and trust policy
335
+ - agent / longagent behavior
336
+ - usage and budget limits
337
+ - UI / theme settings
338
+ - MCP and extension loading
339
+
340
+ ### Project structure / 项目结构
341
+ - `src/repl.mjs` — main REPL assembly surface
342
+ - `src/repl/` extracted REPL seams
343
+ - `src/ui/` REPL panels and render helpers
344
+ - `src/session/` execution loop, memory, recovery, prompts
345
+ - `src/orchestration/` background and longagent orchestration
346
+ - `src/skill/`, `src/plugin/`, `src/mcp/` extension systems
347
+
348
+ **Useful docs / 推荐文档**
349
+ - [Example config](docs/config.example.yaml)
350
+ - [REPL roadmap](docs/repl-roadmap-0.1.27-0.1.36.md)
351
+
352
+ ---
353
+
354
+ <a id="release-status"></a>
355
+ ## Release Status / 发布状态
356
+
357
+ **Current stable / 当前稳定版本**: `v0.2.1`
358
+ **Latest releases / 最新发布**: [GitHub Releases](https://github.com/kkelly-offical/kkcode/releases)
359
+ **Package / 包地址**: [npm](https://www.npmjs.com/package/@kkelly-offical/kkcode)
360
+
361
+ **English**
362
+ - `0.2.1` rebuilds kkcode around Assistant as the default general-purpose lane, with dedicated Agent and LongAgent modes for coding work.
363
+ - The public Ask lane has been removed; question, explanation, and lightweight personal-assistant tasks now route to Assistant.
364
+
365
+ **中文**
366
+ - `0.2.1` 将 kkcode 重构为以 Assistant 为默认入口的通用个人助手,同时保留专门面向代码工作的 Agent 和 LongAgent 模式。
367
+ - 公共 Ask 通道已移除;问答、解释和轻量个人助手任务现在统一路由到 Assistant。
368
+
369
+ ---
370
+
371
+ <a id="compatibility-limits-and-roadmap"></a>
372
+ ## Compatibility, Limits & Roadmap / 兼容性、边界与路线图
373
+
374
+ **What this README does claim / 本 README 明确声明的能力**
375
+ - terminal-native coding workflows
376
+ - governed execution and permissions
377
+ - staged LongAgent orchestration
378
+ - MCP and local extension surfaces
379
+ - session/background/task visibility
380
+
381
+ **What this README does not promise / 本 README 不承诺的能力**
382
+ - GUI-first product workflows
383
+ - IDE-native UX parity
384
+ - desktop automation platform behavior
385
+ - marketplace-style plugin ecosystem
386
+
387
+ **Roadmap references / 路线图参考**
388
+ - [REPL roadmap 0.1.27 → 0.1.36](docs/repl-roadmap-0.1.27-0.1.36.md)
389
+ - [kkcode vs claudenext compatibility notes](docs/kkcode-vs-claudenext-private-agent-longagent-compat.md)
390
+ - [kkcode vs claudenext report](docs/kkcode-vs-claudenext-private-agent-longagent-report.md)
391
+
392
+ ---
393
+
394
+ <a id="faq"></a>
395
+ ## FAQ / 常见问题
396
+
397
+ **Q: When should I use `longagent`? / 什么时候该用 `longagent`?**
398
+ A: Use it when the task is clearly multi-stage, cross-file, or needs ownership/gates. Ordinary terminal assistance should stay in `assistant`; small coding inspect/patch/verify loops should stay in `agent` / `code`.
399
+
400
+ **Q: Can kkcode work with multiple providers? / kkcode 支持多模型厂商吗?**
401
+ A: Yes. Provider switching is built into config and the REPL command surface.
402
+
403
+ **Q: Can I extend kkcode locally? / 可以本地扩展吗?**
404
+ A: Yes. Commands, skills, hooks, tools, agents, and plugin manifests all have local-first support.
405
+
406
+ **Q: Does kkcode promise GUI or IDE parity? / 是否承诺 GUI 或 IDE 对等体验?**
407
+ A: No. This release line is CLI-first and does not overclaim GUI-first capability.
408
+
409
+ ---
410
+
411
+ <a id="contributing"></a>
412
+ ## Contributing / 贡献
413
+
414
+ **English**
415
+ - Keep changes small, testable, and reviewable.
416
+ - Run validation before pushing:
417
+ - `npm run lint`
418
+ - `npm run typecheck`
419
+ - `node ./scripts/run-node-tests.mjs`
420
+ - `npm run release:verify`
421
+
422
+ **中文**
423
+ - 贡献尽量保持小步、可验证、可审阅。
424
+ - 推送前建议至少运行:
425
+ - `npm run lint`
426
+ - `npm run typecheck`
427
+ - `node ./scripts/run-node-tests.mjs`
428
+ - `npm run release:verify`
429
+
430
+ 欢迎中英双语 issue / PR。
431
+
432
+ ---
433
+
434
+ <a id="license"></a>
435
+ ## License / 许可证
436
+
437
+ kkcode is licensed under **GPL-3.0**.
438
+ See [LICENSE](LICENSE) for the full text.
439
+
440
+ ---
441
+
442
+ <a id="further-reading"></a>
443
+ ## Further Reading / 延伸阅读
444
+
445
+ - [CLI General Assistant Capability Matrix](docs/cli-general-assistant-capability-matrix.md)
446
+ - [0.1.13 Mode Lane Contract](docs/kkcode-0.1.13-mode-lane-contract.md)
447
+ - [Task Delegation Contract Matrix](docs/task-delegation-contract-matrix.md)
448
+ - [Agent / LongAgent Extension Guide](docs/agent-longagent-compat-extension-guide.md)
449
+ - [ClaudeNext Agent / LongAgent Skills Compatibility](docs/claudenext-agent-longagent-skills-compat.md)
450
+ - [REPL roadmap 0.1.27 → 0.1.36](docs/repl-roadmap-0.1.27-0.1.36.md)
451
+ - [Git automation usage](docs/GIT_AUTO_USAGE.md)
452
+ - [Edit diagnostics feedback contract](docs/edit-diagnostics-feedback-contract.md)