@klarkxy/dsh-model-hub 0.1.2 → 0.1.4

package/README.md

@@ -4,7 +4,7 @@
 
 [简体中文](docs/README.zh-CN.md)
 
-See and change the models used by your DeepSeek Harness plugins from one page, and manage which models subagents may use.
+See and change the models used by your DeepSeek Harness plugins from one page, manage which models subagents may use, and watch which model and reasoning effort each running subagent is on.
 
 ## Install
 
@@ -18,13 +18,17 @@ Replace `web` with your profile name and restart the profile.
 
 Open **Plugins → Model Hub**.
 
-- **Fields**: every model setting exposed by installed plugins, such as Auto Title, Git Commit, Safe Auto, and ChatGPT. Search by plugin, purpose, or current model, then pick a new model. Changes are saved by the plugin that owns the setting.
-- **Subagent Cast**: turn the subagent model allowlist on or off and add, replace, or remove models. You can give each model a short task tendency; the main agent reads it as a hint when choosing a subagent model, not as a rule.
+- **Fields**: every model setting exposed by installed plugins, such as Auto Title, Git Commit, and Safe Auto. Search by plugin, purpose, or current model, then pick a new model. Changes are saved by the plugin that owns the setting.
+- **Subagent Cast**: turn the subagent model allowlist on or off and add, replace, or remove models. When enabled with at least one model, it guides the agent to discover the models available in the current conversation and explicitly choose one when delegating. For independent reviews and alternative approaches, it encourages a different model when suitable, while allowing the same model when it fits better. You can add your own task tendencies for individual models; they remain optional hints, and no model-specific tendencies are supplied by default. Below the list, a live view shows each running subagent with the model and reasoning effort it is currently using; an effort marked as adapter default comes from the adapter rather than an explicit choice.
 
 ## Limits
 
 - Only plugins that expose model settings appear.
-- If a setting changed elsewhere, refresh and choose again.
+- Changes made elsewhere update the page automatically. If an edit conflicts with a newer change, refresh and choose again.
+- If the model list cannot load, use **Retry**. The last available list stays visible; warnings identify providers that could not be loaded while other providers remain usable.
+- Model selection guidance requires host tools that support choosing a subagent model. Existing conversations retain their saved model lists; start a new conversation to use an updated list.
+- Guidance does not force a model choice or add extra subtasks solely for variety.
+- Remote subagents and finished runs have no readable route; the live list covers only live local subagents.
 - No usage statistics.
 
 ## License

package/docs/README.zh-CN.md

@@ -4,7 +4,7 @@
 
 [English](../README.md)
 
-在一个页面查看并修改各个 DeepSeek Harness 插件使用的模型，也可以管理子代理可用的模型。
+在一个页面查看并修改各个 DeepSeek Harness 插件使用的模型，管理子代理可用的模型，并查看运行中子代理当前使用的模型与思考强度。
 
 ## 安装
 
@@ -18,13 +18,17 @@ dsh plugin --profile web add @klarkxy/dsh-model-hub
 
 打开「插件 → 模型枢纽」。
 
-- **字段总览**：列出已安装插件提供的模型设置，例如自动标题、Git 提交、安全自动和 ChatGPT。可以按插件、用途或当前模型搜索，再选择新模型；改动由对应插件保存。
-- **子代理阵容**：开关子代理模型白名单，添加、替换或移除模型。每个模型可以写一句「任务倾向」，主智能体选择子代理模型时会参考它，但不强制。
+- **字段总览**：列出已安装插件提供的模型设置，例如自动标题、Git 提交和安全自动。可以按插件、用途或当前模型搜索，再选择新模型；改动由对应插件保存。
+- **子代理阵容**：开关子代理模型白名单，添加、替换或移除模型。开启且至少有一个模型时，会引导智能体发现当前会话可用的模型，并在派发时显式选择。独立复核和方案对比会优先考虑合适的不同模型，同模型更适合时仍可使用。你可以自行填写各模型的「任务倾向」作为可选参考；插件不预置模型专用倾向。页面下方实时列出运行中的子代理及各自当前使用的模型和思考强度；标注「适配器默认」的强度来自适配器而非显式选择。
 
 ## 限制
 
 - 只显示提供了模型设置的插件。
-- 设置在别处被改动时，刷新后重新选择。
+- 设置在别处改动时，页面会自动更新。编辑与较新的改动冲突时，请刷新后重新选择。
+- 模型列表读取失败时，点击「重试」。上次可用的列表会保留；部分提供方读取失败时会显示提示，其他提供方仍可使用。
+- 选模引导需要宿主工具支持选择子代理模型。已有会话保留各自的模型名单；修改名单后，新建会话才能使用更新后的名单。
+- 引导不强制模型选择，也不要求仅为增加模型种类而新增子任务。
+- 远程子代理和已结束的子代理没有可读取的路由；运行列表只覆盖存活的本地子代理。
 - 不提供用量统计。
 
 ## 许可证