ai_co_discuss 0.2.73 → 0.2.74

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 (6) hide show
  1. package/README.md +12 -0
  2. package/README.zh-CN.md +13 -0
  3. package/docs/architecture.md +2 -0
  4. package/package.json +1 -1
  5. package/web/modern.css +1 -1
  6. package/web/modern.js +393 -397
package/README.md CHANGED
@@ -26,12 +26,23 @@ Most multi-agent frameworks spin up fresh agents inside their own runtime. AI Co
26
26
 
27
27
  - **Delegate by mention** — `@` an agent and it works in its own original session; several can run in parallel. Unaddressed messages go to an AI host that routes the work.
28
28
  - **Structured discussions** — set a goal and round / time limits; the host schedules speakers and follow-ups, then the room returns to free collaboration with history and conclusions kept.
29
+ - **Smart Scheduling** — optionally let a decision model handle routine routing instead of waking the host.
29
30
  - **Mix any participant** — local CLIs and desktop apps, model APIs, manually connected sessions and web chatbots in one room.
30
31
  - **Shared workspace** — point a room at a project directory and grant each local agent read-only or writable access.
31
32
  - **Live roster** — invite, pause, recall or remove members at any time, including resuming an existing agent session.
32
33
  - **Portable records** — attachments, ZIP export / import between sites, and sourced reports via the `discuss-report` skill.
33
34
  - **Local first** — runs entirely on your machine; optionally host the site on Cloudflare or Docker while agents stay local.
34
35
 
36
+ ## Smart Scheduling
37
+
38
+ Many turns in a room are routing decisions: who a message is for, whether members should answer together or in order, whether a round should continue. With Smart Scheduling enabled, these are answered by a decision model (TypeSafe Jev; OpenAI Decisions is experimental) instead of waking the AI host, typically in about a second.
39
+
40
+ - Unaddressed messages go to the member they are clearly meant for, to everyone, or to no one; otherwise to the host.
41
+ - In discussions, it resumes the queue after an interjection and picks the next speaker between rounds.
42
+ - It acts only above a confidence threshold, and falls back to the host on low confidence or errors. Each decision is shown in the conversation and can be rated.
43
+
44
+ It is opt-in per room and uses your own key; the room title and recent messages are sent to the provider. See [Smart Scheduling](docs/runner.md#智能调度可选).
45
+
35
46
  ## Supported Participants
36
47
 
37
48
  | Mode | Participants | Notes |
@@ -77,6 +88,7 @@ flowchart LR
77
88
  S -- events --> R[Local runner<br>codiscuss service]
78
89
  R -- wake / resume --> A[Claude Code · Codex · Antigravity]
79
90
  R -- API calls --> M[Model APIs]
91
+ R -. routing, optional .-> D[Decision model]
80
92
  A -- codiscuss CLI --> S
81
93
  M -- room tools --> S
82
94
  W[Web chatbots] -. read-only page .-> S
package/README.zh-CN.md CHANGED
@@ -26,12 +26,23 @@
26
26
 
27
27
  - **点名分工**：`@` 哪位 AI，它就在自己的原会话中处理，多位可并行；未点名的消息由 AI 主持人分派。
28
28
  - **结构化讨论**：设定目标与轮次 / 时长上限，主持人安排发言与追问；结束后回到自由协作，记录与结论保留。
29
+ - **智能调度**：可选由决策模型处理日常分派，不必每次唤醒主持人。
29
30
  - **混合接入**：本地 CLI 与桌面 App、模型 API、手动接入的会话和在线网页 AI 可同房协作。
30
31
  - **共享项目**：为房间指定项目目录，并为每位本地 AI 设置只读或可写。
31
32
  - **随时调整成员**：随时邀请、暂离、请回或移除成员，也可以续接一个已有的 AI 会话。
32
33
  - **可迁移记录**：附件、跨站点 ZIP 导出 / 导入，并可用 `discuss-report` 技能生成带来源的报告。
33
34
  - **本机优先**：完全在本机运行；也可将网站部署到 Cloudflare 或 Docker，AI 仍在本机执行。
34
35
 
36
+ ## 智能调度
37
+
38
+ 房间里的许多轮次只是分派：消息是对谁说的、成员同时还是依次回答、讨论是否继续下一轮。开启智能调度后，这些判断由决策模型（TypeSafe Jev；OpenAI Decisions 为实验性）完成，不必唤醒 AI 主持人，通常约一秒。
39
+
40
+ - 未点名的消息：明确对某位成员说时直接交给他，或请全体回答、无需回应；其余仍交给主持人。
41
+ - 讨论中：用户插话后继续原有安排，轮次之间选出下一位发言人。
42
+ - 只在置信度达到阈值时执行，把握不足或出错时交回主持人；每次判断都显示在对话中，可评价。
43
+
44
+ 按房间开启，使用你自己的密钥；房间标题与最近消息会发给所选供应商。详见 [智能调度](docs/runner.md#智能调度可选)。
45
+
35
46
  ## 支持的参与者
36
47
 
37
48
  | 接入方式 | 参与者 | 说明 |
@@ -74,6 +85,7 @@ flowchart LR
74
85
  S -- 事件 --> R[本机接入服务<br>codiscuss service]
75
86
  R -- 唤醒 / 续接 --> A[Claude Code · Codex · Antigravity]
76
87
  R -- API 调用 --> M[模型 API]
88
+ R -. 调度判断（可选） .-> D[决策模型]
77
89
  A -- codiscuss CLI --> S
78
90
  M -- 房间工具 --> S
79
91
  W[网页 AI] -. 只读阅读页 .-> S
@@ -88,6 +100,7 @@ flowchart LR
88
100
  ## 更多入口
89
101
 
90
102
  - **线上网站 + 本机 AI**：在运行 AI 的电脑上执行一次 `codiscuss local connect https://你的站点`。见 [线上网页连接本机](docs/runner.md#线上网页连接本机)。
103
+ - **手机或平板建房**：在电脑上连接一次并保存常用项目后，任何设备登录同一网站的管理员账号即可新建房间，由那台电脑唤醒 AI，项目路径只留在电脑上。见 [其他设备经已配对电脑建房](docs/runner.md#其他设备经已配对电脑建房)。
91
104
  - **从 AI 对话发起**：让本地 AI 使用 `discuss-host` 和 `discuss-participant` 技能建房并主持。见 [从 AI 对话发起](docs/runner.md#从-ai-对话发起)。
92
105
  - **模型 API**：按成员配置供应商、Base URL、请求协议与模型。见 [API 接入](docs/runner.md#api-接入)。
93
106
 
package/docs/architecture.md CHANGED
@@ -145,6 +145,8 @@ Python 在一个只读事务快照中查询；D1 使用同一个 batch 事务读
145
145
 
146
146
  `frontend/` 保存 Vue、Markdown 和备份源码；`web/` 保存共享控制器、HTML 入口、图标和构建产物。生成产物随仓库提供，保证仅安装 Python 的用户也能运行网页；不在生成文件中手工修改功能。
147
147
 
148
+ 新建房间表单须至少选中一位成员，决策模型不计入成员；创建按钮与表单提交（含 Enter）共用就绪校验，防止键盘绕过禁用状态。成员添加行紧跟“谁来参与”标题，接入方式由卡片上的“程序启动 / 手动启动”选项表达。
149
+
148
150
  | 源码 | 职责 |
149
151
  | --- | --- |
150
152
  | `main.js`、`App.vue`、`ChannelWorkspace.vue` | 协作频道入口、主题与整体布局 |
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "ai_co_discuss",
3
- "version": "0.2.73",
3
+ "version": "0.2.74",
4
4
  "bin": {
5
5
  "codiscuss": "runner/codiscuss.mjs"
6
6
  },