@ycodium-ai/plugin-im 0.2.3250 → 0.2.3254

package/README.md

@@ -260,9 +260,9 @@ is sent to the chat. `chat_type` (1 single, 2 group) comes from the last inbound
 conversation and is left out when none has arrived since the plugin started, as plain text does.
 A picture goes as `image` only when it is PNG, JPEG or GIF within 10 MB (`imageMimeTypes`).
 
-Chat commands: `/stop`, `/continue`, `/status`, `/new`, `/sessions`, `/resume`, `/history [count]`, `/model`,
-`/plan`, `/repo`, `/group`, `/team`, `/thread`, `/help`, `/panel`. In a chat with no project
-`/sessions`, `/resume` and the team commands answer with the project picker. `/sessions` and `/resume` only list and
+Chat commands: `/stop`, `/continue`, `/status`, `/version`, `/new`, `/sessions`, `/resume`, `/history [count]`, `/model`,
+`/reasoning`, `/plan`, `/repo`, `/group`, `/team`, `/thread`, `/help`, `/panel`. In a chat with no project
+`/sessions`, `/resume` and the team commands answer with the project picker. `/version` still answers. `/sessions` and `/resume` only list and
 switch threads this plugin opened, since the host lets a plugin send only into
 its own threads. `/continue` calls `threads.turnQueue.resume` for the chat's thread
 when its queue is paused (a `/stop` or a failure pauses it), which is how WeCom
@@ -270,8 +270,12 @@ resumes a paused queue from the chat; it is not `/resume`, which already switche
 
 `/history [count]` previews the current session's recent user and assistant messages in private chats only (default 3, maximum 5), oldest first. It does not start a turn or change the session. Images and files appear as placeholders; long text is split for the platform, with a notice when the preview text is incomplete. Type the command in the chat; panels and native command menus do not list it. With the reply language set to Chinese, `会话历史 [数量]` works too.
 
+`/version` answers with one line, the plugin name and its version. It takes no arguments; if anything follows it, the reply is a usage reminder. It works in private chats and groups, does not read the session, and does not start a turn. Type it; panels and native command menus do not list it. There is no Chinese alias.
+
 On Lark, Telegram, and the WeCom smart bot, `/panel` sends a command panel, and `/help` stays text and ends with a pointer to it. WeCom App does not implement the panel surface, so `/panel` answers with the text list and `/help` does not mention `/panel`. Lark cards and Telegram inline keyboards show all command buttons. The WeCom smart bot uses a template card from `src/wecomPanel.ts`; its main menu is limited to six buttons. Commands that do not fit are listed in `/help` and can be typed into the chat. Commands that take an argument (a thread, a project, a model) open a second view on the same message with a Back button. These lists let you reach every option by paging: Lark and Telegram show eight items per page, and WeCom shows three, leaving room for Back, Previous, and Next. A tap on the WeCom card updates that same card; when the update cannot be applied, `src/panelController.ts` posts a new card and the old one stops accepting taps. A busy placeholder is not drawn on that card: it stays as it is until the next interactive view. Telegram buttons carry only a short `pn:<version>:<index>` token (`callback_data` is capped at 64 bytes); the action stays in memory and a tap on an older keyboard version counts as expired. Each card keeps its own view stack in memory (`src/panelController.ts`), so after a restart an old card only answers that it has expired. A tap runs the same handler as the typed command and waits in the same per-chat queue as typed messages. Another platform opts in by implementing the optional `panel` surface on its adapter (`src/adapter.ts`).
 
+`/reasoning` shows the current model's reasoning efforts; `/reasoning <value>` or `/reasoning <number>` switches it. Numbers follow the available values at the time you send the command, so send `/reasoning` first to check the choices. See [Model and Plan Mode](../../docs/user/im-bots.md#model-and-plan-mode) for details.
+
 `/team` uses the chat's selected project. It reads the roster from the `team` plugin and
 each text message invokes `startGoal` with a stable request ID derived from the bot,
 platform, chat, and inbound message ID. A successful goal binds the chat to the lead's

package/dist/index.js

@@ -126699,7 +126699,18 @@ var readChatPatch = (data) => {
   else if (isRecord(data["model"])) {
     const instanceId = nonEmpty3(data["model"]["instanceId"]);
     const model = nonEmpty3(data["model"]["model"]);
-    if (instanceId !== null && model !== null) patch.model = { instanceId, model };
+    if (instanceId !== null && model !== null) {
+      const options = data["model"]["options"];
+      patch.model = {
+        instanceId,
+        model,
+        ...Array.isArray(options) ? {
+          options: options.flatMap(
+            (entry) => isRecord(entry) && nonEmpty3(entry["id"]) !== null && (typeof entry["value"] === "string" || typeof entry["value"] === "boolean") ? [{ id: String(entry["id"]), value: entry["value"] }] : []
+          )
+        } : {}
+      };
+    }
   }
   if (data["preset"] === null) patch.preset = null;
   else if (nonEmpty3(data["preset"]) !== null) patch.preset = nonEmpty3(data["preset"]);
@@ -128340,7 +128351,7 @@ var makeChatSetup = (core, deps) => {
 };
 
 // src/commands.ts
-import { threadOrNull as threadOrNull2 } from "@ycodium-ai/plugin-api";
+import { threadOrNull as threadOrNull3 } from "@ycodium-ai/plugin-api";
 
 // ../../node_modules/.pnpm/effect@4.0.0-beta.103_patch_hash=78523ded7c3dfc3efd3cec0247f4960d00eaa3d3c96a2077457694a4d8b453c6/node_modules/effect/dist/Pipeable.js
 var pipeArguments = (self2, args2) => {
@@ -133557,7 +133568,10 @@ var PlanCommandMode = Literals(["toggle", "on", "off", "invalid"]);
 var ParsedChatCommand = Union2([
   SimpleChatCommand,
   Struct({ kind: Literal2("sessions"), arg: optionalKey2(String4) }),
-  Struct({ kind: Literals(["resume", "model", "repo"]), arg: String4 }),
+  Struct({
+    kind: Literals(["resume", "model", "reasoning", "repo", "version"]),
+    arg: String4
+  }),
   Struct({ kind: Literal2("history"), count: NullOr(Number5) }),
   Struct({
     kind: Literal2("preset"),
@@ -133603,6 +133617,8 @@ var parseChatCommand = (input) => {
   if (/^\/stop$/i.test(trimmed)) return { kind: "stop" };
   if (/^\/continue$/i.test(trimmed)) return { kind: "continue" };
   if (/^\/status$/i.test(trimmed)) return { kind: "status" };
+  const versionArg = trimmed.match(/^\/version(?:\s+([\s\S]*))?$/iu);
+  if (versionArg !== null) return { kind: "version", arg: (versionArg[1] ?? "").trim() };
   if (/^\/new$/i.test(trimmed)) return { kind: "new" };
   if (/^\/help$/i.test(trimmed)) return { kind: "help" };
   if (/^\/panel$/i.test(trimmed)) return { kind: "panel" };
@@ -133626,6 +133642,13 @@ var parseChatCommand = (input) => {
   }
   const model = modelArgument(input.text, input.chinese);
   if (model !== void 0) return { kind: "model", arg: model };
+  const reasoning = commandArgument(
+    trimmed,
+    input.chinese,
+    /^\/reasoning(?:\s+([\s\S]*))?$/i,
+    /^思考强度(?:\s+([\s\S]*))?$/
+  );
+  if (reasoning !== void 0) return { kind: "reasoning", arg: reasoning };
   const preset = trimmed.match(/^\/preset(?:\s+([\s\S]*))?$/iu);
   if (preset !== null)
     return {
@@ -133656,12 +133679,14 @@ var helpText = (t, options = {}) => [
   t("help.stop"),
   t("help.continue"),
   t("help.status"),
+  t("help.version"),
   t("help.new"),
   t("help.sessions"),
   t("help.resume"),
   t("help.history", HISTORY_COUNTS),
   t("help.syncNote"),
   t("help.model"),
+  t("help.reasoning"),
   t("help.preset"),
   t("help.plan"),
   t("help.repo"),
@@ -133841,6 +133866,276 @@ ${body}`);
   };
 };
 
+// src/reasoningCommand.ts
+import { modelSelectionStringOption as modelSelectionStringOption2 } from "@ycodium-ai/plugin-api/executor";
+
+// src/currentModel.ts
+import { threadOrNull as threadOrNull2 } from "@ycodium-ai/plugin-api";
+
+// src/botModel.ts
+var resolveDefaultModel = async (bot, chatModel, projectDefault, providers, knownInstances) => {
+  if (chatModel !== null) return chatModel;
+  if (bot.defaultModel !== void 0) {
+    const instances = knownInstances === void 0 ? await providers.list({}).catch(() => null) : knownInstances;
+    if (!isModelCatalogComplete(instances) || modelCatalogHas(modelChoices(instances ?? []), bot.defaultModel)) {
+      return bot.defaultModel;
+    }
+  }
+  return projectDefault;
+};
+
+// src/threadUsable.ts
+var threadUsable = (binding, thread) => thread !== null && thread.archivedAt == null && thread.projectId === binding.projectId;
+
+// src/currentModel.ts
+var makeCurrentModelResolver = (core) => {
+  const { host } = core;
+  return async (binding, {
+    instances,
+    onReadError = (error) => {
+      throw error;
+    }
+  } = {}) => {
+    const chat = chatOf(core.state, chatKeyOf(binding));
+    if (chat.threadId !== null) {
+      const thread = await threadOrNull2(host.threads, chat.threadId, { archived: "include" }).catch(
+        onReadError
+      );
+      if (threadUsable(binding, thread))
+        return { selection: thread.modelSelection, threadId: chat.threadId };
+    }
+    if (chat.model !== null) return { selection: chat.model, threadId: null };
+    const project = binding.projectId === "" ? null : await host.projects.get({ projectId: binding.projectId }).catch(onReadError);
+    return {
+      selection: await resolveDefaultModel(
+        binding,
+        chat.model,
+        project?.defaultModelSelection ?? null,
+        host.providers.instances,
+        instances
+      ),
+      threadId: null
+    };
+  };
+};
+var saveModelSelection = async (core, binding, threadId, model) => {
+  if (threadId !== null) {
+    try {
+      await core.host.threads.update({
+        requestId: crypto.randomUUID(),
+        threadId,
+        modelSelection: model
+      });
+    } catch (error) {
+      core.host.log.warn("chat could not switch the thread model", {
+        error: describeError(error)
+      });
+      return false;
+    }
+  }
+  core.commit("chat-updated", { chatKey: chatKeyOf(binding), model });
+  return true;
+};
+
+// src/reasoningOption.ts
+import { modelOptionLabelKind, modelSelectionStringOption } from "@ycodium-ai/plugin-api/executor";
+var isReasoningOption = (id) => modelOptionLabelKind(id) !== void 0;
+var reasoningOption = (model) => model.options.find((option2) => isReasoningOption(option2.id) && option2.values.length > 0);
+var selectedReasoning = (selection) => {
+  const option2 = selection?.options?.find((entry) => isReasoningOption(entry.id));
+  if (option2 === void 0) return void 0;
+  const value2 = modelSelectionStringOption(selection, option2.id);
+  return value2 === void 0 ? void 0 : { id: option2.id, value: value2 };
+};
+
+// src/reasoningCommand.ts
+var isDefaultArgument = (arg, chinese) => arg.toLowerCase() === "default" || chinese && arg === "\u9ED8\u8BA4";
+var makeReasoningCommand = (core, currentModel) => {
+  const { host, t, deliver } = core;
+  const readOptions = async (binding) => {
+    const resolved = await currentModel(binding);
+    const selection = resolved.selection;
+    const catalog = selection === null ? null : await host.providers.models.list({ instanceId: selection.instanceId });
+    const model = catalog?.models.find((entry) => entry.slug === selection?.model);
+    return { ...resolved, option: model === void 0 ? void 0 : reasoningOption(model) };
+  };
+  return async (binding, arg) => {
+    const resolved = await readOptions(binding).catch(() => null);
+    if (resolved === null) {
+      host.log.warn("chat could not read reasoning options");
+      await deliver(binding, t("reasoning.readFailed"));
+      return;
+    }
+    const { selection, threadId, option: option2 } = resolved;
+    if (selection === null) {
+      await deliver(binding, t("reasoning.noModel"));
+      return;
+    }
+    const wanted = arg.trim();
+    if (isDefaultArgument(wanted, t.language === "zh-CN")) {
+      if (selectedReasoning(selection) === void 0) {
+        await deliver(binding, t(threadId === null ? "reasoning.resetNext" : "reasoning.resetNow"));
+        return;
+      }
+      const next2 = {
+        ...selection,
+        options: (selection.options ?? []).filter((entry) => !isReasoningOption(entry.id))
+      };
+      if (!await saveModelSelection(core, binding, threadId, next2)) {
+        await deliver(binding, t("reasoning.setFailed"));
+        return;
+      }
+      await deliver(binding, t(threadId === null ? "reasoning.resetNext" : "reasoning.resetNow"));
+      return;
+    }
+    if (option2 === void 0) {
+      await deliver(binding, t("reasoning.unsupported"));
+      return;
+    }
+    if (arg === "") {
+      const current = modelSelectionStringOption2(selection, option2.id);
+      const listed = current !== void 0 && option2.values.includes(current);
+      const title = current === void 0 ? "reasoning.listTitleDefault" : listed ? "reasoning.listTitle" : "reasoning.listTitlePlain";
+      const text3 = [
+        ...current === void 0 ? [] : [t(listed ? "reasoning.current" : "reasoning.currentUnlisted", { value: current })],
+        ...formatChatChoices(
+          option2.values,
+          (entry) => entry,
+          (entry) => entry === current,
+          {
+            title: t(title),
+            more: (count) => t("reasoning.more", { count }),
+            footer: t("reasoning.usage")
+          }
+        )
+      ].join("\n").trim();
+      await deliver(binding, text3);
+      return;
+    }
+    const value2 = option2.values.find((entry) => entry === wanted) ?? pickChatChoice(wanted, option2.values, (entry, needle) => entry.toLowerCase() === needle);
+    if (value2 === null) {
+      const shown = option2.values.slice(0, CHAT_CHOICE_LIST_MAX);
+      await deliver(
+        binding,
+        [
+          /^\d+$/.test(wanted) ? t("reasoning.numberOutOfRange", { number: wanted, max: shown.length }) : t("reasoning.notFound", { query: arg }),
+          t("reasoning.usage"),
+          t("reasoning.choices", { values: shown.join(" / ") }),
+          ...option2.values.length > shown.length ? [t("reasoning.more", { count: option2.values.length - shown.length })] : []
+        ].join("\n")
+      );
+      return;
+    }
+    const next = {
+      ...selection,
+      options: [
+        ...(selection.options ?? []).filter((entry) => !isReasoningOption(entry.id)),
+        { id: option2.id, value: value2 }
+      ]
+    };
+    if (!await saveModelSelection(core, binding, threadId, next)) {
+      await deliver(binding, t("reasoning.setFailed"));
+      return;
+    }
+    await deliver(
+      binding,
+      t(threadId === null ? "reasoning.setNext" : "reasoning.setNow", { value: value2 })
+    );
+  };
+};
+
+// package.json
+var package_default = {
+  name: "@ycodium-ai/plugin-im",
+  version: "0.2.3254",
+  description: "Talk to your agents from Telegram, WeCom, Lark, or WeChat (experimental).",
+  keywords: [
+    "im",
+    "lark",
+    "telegram",
+    "wechat",
+    "wecom",
+    "wecom-app",
+    "ycodium",
+    "ycodium-plugin"
+  ],
+  license: "UNLICENSED",
+  files: [
+    "dist",
+    "assets",
+    "locales"
+  ],
+  type: "module",
+  publishConfig: {
+    access: "public"
+  },
+  scripts: {
+    build: "node build.mjs",
+    prepack: "pnpm run build",
+    typecheck: "tsgo --noEmit && tsgo --noEmit -p tsconfig.ui.json",
+    test: "vp test run"
+  },
+  devDependencies: {
+    "@effect/vitest": "catalog:",
+    "@larksuiteoapi/node-sdk": "^1.73.0",
+    "@types/markdown-it": "^14.2.0",
+    "@types/react": "~19.2.14",
+    "@ycodium-ai/plugin-api": "workspace:*",
+    effect: "catalog:",
+    esbuild: "catalog:",
+    "markdown-it": "^15.0.2",
+    react: "19.2.6",
+    "vite-plus": "catalog:",
+    "ycodium-react-test-harness": "workspace:*"
+  },
+  ycodium: {
+    plugin: "./dist/index.js",
+    displayName: "Chat bridge",
+    description: "Talk to your agents from Telegram, WeCom, Lark, or WeChat (experimental).",
+    icon: "messageCircle",
+    permissions: [
+      {
+        kind: "sessionProjection"
+      },
+      {
+        kind: "threads.user"
+      },
+      {
+        kind: "workspace.read"
+      },
+      {
+        kind: "commands.dispatch",
+        commands: [
+          "thread.create",
+          "thread.meta.update",
+          "thread.external-messages.append",
+          "thread.turn.start",
+          "thread.turn.interrupt",
+          "thread.approval.respond",
+          "thread.user-input.respond",
+          "thread.turn-intent.run",
+          "thread.turn-intent.cancel",
+          "thread.turn-queue.resume"
+        ]
+      },
+      {
+        kind: "ui.seats",
+        seats: [
+          "settings.section"
+        ]
+      }
+    ],
+    ui: {
+      assets: "./assets",
+      entry: "ui.js"
+    }
+  }
+};
+
+// src/versionCommand.ts
+var pluginVersion = package_default.version;
+var versionReply = (t, arg) => arg === "" ? t("version.current", { version: pluginVersion }) : t("version.usage");
+
 // src/commands.ts
 var NEEDS_PROJECT_COMMANDS = /* @__PURE__ */ new Set(["sessions", "resume", "selectTeam", "selectThread"]);
 var makeCommands = (core, deps) => {
@@ -133851,6 +134146,7 @@ var makeCommands = (core, deps) => {
   const { handleModel, handlePlan } = deps.modelPlan;
   const { openPanel } = deps;
   const handleHistory = makeHistoryCommand(core);
+  const handleReasoning = makeReasoningCommand(core, deps.modelPlan.resolveCurrentModel);
   const targetOf = (binding) => chatOf(core.state, chatKeyOf(binding)).target;
   const executeCommand = async (binding, command, senderId = null, chatKind = void 0) => {
     const chatKey = chatKeyOf(binding);
@@ -133912,6 +134208,9 @@ var makeCommands = (core, deps) => {
       case "model":
         await handleModel(binding, command.arg);
         return;
+      case "reasoning":
+        await handleReasoning(binding, command.arg);
+        return;
       case "preset":
         await deps.handlePreset(binding, command);
         return;
@@ -133922,6 +134221,9 @@ var makeCommands = (core, deps) => {
         bindChat(chatKey, null);
         await deliver(binding, t("command.newSession"));
         return;
+      case "version":
+        await deliver(binding, versionReply(t, command.arg));
+        return;
       case "status": {
         const threadId = chatOf(core.state, chatKey).threadId ?? void 0;
         const routed = targetOf(binding);
@@ -133939,7 +134241,7 @@ var makeCommands = (core, deps) => {
           );
           return;
         }
-        const thread = await threadOrNull2(host.threads, threadId);
+        const thread = await threadOrNull3(host.threads, threadId);
         const running = isSessionRunning(thread?.session?.status);
         await deliver(
           binding,
@@ -133989,7 +134291,7 @@ var makeCommands = (core, deps) => {
           await deliver(binding, t("command.stopNoSession"));
           return;
         }
-        const thread = await threadOrNull2(host.threads, threadId);
+        const thread = await threadOrNull3(host.threads, threadId);
         const activeTurnId = thread?.session?.activeTurnId ?? null;
         if (activeTurnId === null) {
           await deliver(binding, t("command.stopNoTurn"));
@@ -135428,7 +135730,7 @@ var makeImCore = (input) => {
 };
 
 // src/conclusionDelivery.ts
-import { threadOrNull as threadOrNull3 } from "@ycodium-ai/plugin-api";
+import { threadOrNull as threadOrNull4 } from "@ycodium-ai/plugin-api";
 
 // src/abortableSleep.ts
 var abortableSleep = (ms, signal) => new Promise((resolve2, reject) => {
@@ -135556,13 +135858,13 @@ ${text3}` : text3;
   const sleep4 = (ms) => abortableSleep(ms, stopped.signal).catch(() => void 0);
   const retryFor = (threadId, owed = ALWAYS_OWED) => ({
     sleep: sleep4,
-    shouldContinue: async () => !disposed && owed() && await threadOrNull3(core.host.threads, threadId) !== null
+    shouldContinue: async () => !disposed && owed() && await threadOrNull4(core.host.threads, threadId) !== null
   });
   return {
     completed,
     retryFor,
     prepareConclusion: async (threadId, turnId, successful) => {
-      const thread = await threadOrNull3(core.host.threads, threadId);
+      const thread = await threadOrNull4(core.host.threads, threadId);
       const final = thread === null ? void 0 : await latestMessage(
         core.host.threads,
         threadId,
@@ -146251,19 +146553,25 @@ var en_default = {
     project: "Project: {{title}}",
     noProject: "No project picked yet. Send /repo to pick one."
   },
+  version: {
+    current: "Ycodium Chat bridge v{{version}}",
+    usage: "Usage: /version (no arguments)."
+  },
   help: {
     title: "Commands:",
     continue: "\xB7 /continue \u2014 resume sending after /stop or a failure paused it; the waiting messages go out",
     stop: "\xB7 /stop \u2014 stop the turn that is running; the session stays",
     status: "\xB7 /status \u2014 what is going on right now",
+    version: "\xB7 /version \u2014 the chat bridge's version (no arguments)",
     new: "\xB7 /new \u2014 forget the current session; the next message starts a new one",
     team: "\xB7 /team \u2014 give this project's team a goal with each message; use /repo to pick the project first",
     thread: "\xB7 /thread \u2014 go back to a normal session",
-    aliases: "Chinese shortcuts (\u505C / \u7EE7\u7EED\u53D1\u9001 / \u72B6\u6001 / \u65B0\u5BF9\u8BDD / \u4F1A\u8BDD\u5217\u8868 / \u5207\u6362\u4F1A\u8BDD / \u4F1A\u8BDD\u5386\u53F2 / \u6362\u6A21\u578B / \u8BA1\u5212\u6A21\u5F0F / \u6362\u9879\u76EE / \u5EFA\u7FA4 / \u6362\u56E2\u961F / \u6362\u5BF9\u8BDD) work when the reply language is Chinese.",
+    aliases: "Chinese shortcuts (\u505C / \u7EE7\u7EED\u53D1\u9001 / \u72B6\u6001 / \u65B0\u5BF9\u8BDD / \u4F1A\u8BDD\u5217\u8868 / \u5207\u6362\u4F1A\u8BDD / \u4F1A\u8BDD\u5386\u53F2 / \u6362\u6A21\u578B / \u601D\u8003\u5F3A\u5EA6 / \u8BA1\u5212\u6A21\u5F0F / \u6362\u9879\u76EE / \u5EFA\u7FA4 / \u6362\u56E2\u961F / \u6362\u5BF9\u8BDD) work when the reply language is Chinese.",
     sessions: "\xB7 /sessions \u2014 list this project's sessions; add a project number from /repo to list the ones this bot opened in another project",
     resume: "\xB7 /resume number-or-id-prefix \u2014 switch sessions; one from another project switches the project too",
     history: "\xB7 /history [count] \u2014 preview recent messages in the current session, in private chats only (default {{defaultCount}}, maximum {{maxCount}})",
     model: "\xB7 /model \u2014 see the model this chat uses; /model number or /model instance/model switches it",
+    reasoning: "\xB7 /reasoning \u2014 see the reasoning effort for this chat's current model; /reasoning value or /reasoning number switches it; /reasoning default resets it to the model default",
     preset: "\xB7 /preset \u2014 see the preset this chat uses for new sessions; /preset number or /preset name-or-id switches it; /preset none clears the chat choice",
     plan: "\xB7 /plan \u2014 toggle plan mode (the agent proposes a plan instead of acting); /plan on, /plan off set it",
     panel: "Send /panel to open the command panel with a button for every command.",
@@ -146271,6 +146579,28 @@ var en_default = {
     group: "\xB7 /group name \u2014 Lark only: the bot creates a group and adds you (bots on other platforms cannot)",
     syncNote: "\xB7 After a private chat attaches a session, web user text and final assistant text sync here (groups never sync)."
   },
+  reasoning: {
+    current: "Current reasoning effort: {{value}}",
+    currentUnlisted: "Current reasoning effort: {{value}} (no longer in the list; it stays in effect until you change it)",
+    resetNow: "Reasoning effort reset to the model default. The current session uses it from the next turn; new sessions in this chat use it too.",
+    resetNext: "Reasoning effort reset to the model default. New sessions in this chat will use it.",
+    notFound: '"{{query}}" is not an available reasoning effort.',
+    numberOutOfRange: "Number {{number}} is out of range; choose 1\u2013{{max}}.",
+    listTitle: "Available reasoning efforts (\u2605 current):",
+    listTitleDefault: "Available reasoning efforts (none chosen; the model default is used):",
+    more: "{{count}} more not shown.",
+    usage: "Usage: /reasoning \u2014 see the reasoning effort for this chat's current model; /reasoning value or /reasoning number switches it; /reasoning default resets it to the model default.",
+    choices: "Choices: {{values}}.",
+    unsupported: "The current model does not support switching the reasoning effort.",
+    readFailed: "Could not read the reasoning options. Try again in a moment.",
+    setFailed: "Switching the reasoning effort failed. Try again in a moment.",
+    setNow: "Reasoning effort set to {{value}}. The current session uses it from the next turn; new sessions in this chat use it too.",
+    setNext: "Reasoning effort set to {{value}}. New sessions in this chat will use it.",
+    cleared: "The new model does not support the previous reasoning effort ({{value}}). It was cleared; the model default will be used.",
+    noModel: "This chat has no model yet. Pick one with /model first.",
+    switchReadFailed: "Could not read the new model's reasoning efforts, so the model was not switched. Try again in a moment.",
+    listTitlePlain: "Available reasoning efforts:"
+  },
   history: {
     usage: "Usage: /history [count] \u2014 1 to {{maxCount}} (larger numbers count as {{maxCount}}), default {{defaultCount}}.",
     directOnly: "In a group this would show the session to everyone. Send /history in a private chat with the bot.",
@@ -147176,19 +147506,25 @@ var zh_CN_default = {
     project: "\u9879\u76EE\uFF1A{{title}}",
     noProject: "\u8FD8\u6CA1\u9009\u9879\u76EE\u3002\u53D1 /repo \u9009\u4E00\u4E2A\u3002"
   },
+  version: {
+    current: "Ycodium \u804A\u5929\u6865 v{{version}}",
+    usage: "\u7528\u6CD5\uFF1A/version\uFF08\u4E0D\u5E26\u53C2\u6570\uFF09\u3002"
+  },
   help: {
     title: "\u80FD\u7528\u7684\u547D\u4EE4\uFF1A",
     continue: "\xB7 /continue \u2014\u2014 /stop \u6216\u51FA\u9519\u6682\u505C\u4E86\u53D1\u9001\u4E4B\u540E\uFF0C\u7528\u5B83\u6062\u590D\uFF0C\u6392\u7740\u7684\u6D88\u606F\u7EE7\u7EED\u53D1\u51FA",
     stop: "\xB7 /stop \u2014\u2014 \u505C\u6389\u6B63\u5728\u8DD1\u7684\u8FD9\u4E00\u8F6E\uFF0C\u4F1A\u8BDD\u7559\u7740",
     status: "\xB7 /status \u2014\u2014 \u73B0\u5728\u5728\u5E72\u4EC0\u4E48",
+    version: "\xB7 /version \u2014\u2014 \u804A\u5929\u6865\u7684\u7248\u672C\uFF08\u4E0D\u5E26\u53C2\u6570\uFF09",
     new: "\xB7 /new \u2014\u2014 \u5FD8\u6389\u5F53\u524D\u4F1A\u8BDD\uFF0C\u4E0B\u4E00\u53E5\u5F00\u65B0\u7684",
     team: "\xB7 /team \u2014\u2014 \u6BCF\u6761\u6D88\u606F\u90FD\u7ED9\u8FD9\u4E2A\u9879\u76EE\u7684\u56E2\u961F\u5F00\u4E00\u4E2A\u76EE\u6807\uFF1B\u5148\u7528 /repo \u9009\u9879\u76EE",
     thread: "\xB7 /thread \u2014\u2014 \u5207\u56DE\u666E\u901A\u4F1A\u8BDD",
-    aliases: "\u4E2D\u6587\u4E5F\u884C\uFF1A\u505C / \u7EE7\u7EED\u53D1\u9001 / \u72B6\u6001 / \u65B0\u5BF9\u8BDD / \u4F1A\u8BDD\u5217\u8868 / \u5207\u6362\u4F1A\u8BDD / \u4F1A\u8BDD\u5386\u53F2 / \u6362\u6A21\u578B / \u8BA1\u5212\u6A21\u5F0F / \u6362\u9879\u76EE / \u5EFA\u7FA4 / \u6362\u56E2\u961F / \u6362\u5BF9\u8BDD\u3002",
+    aliases: "\u4E2D\u6587\u4E5F\u884C\uFF1A\u505C / \u7EE7\u7EED\u53D1\u9001 / \u72B6\u6001 / \u65B0\u5BF9\u8BDD / \u4F1A\u8BDD\u5217\u8868 / \u5207\u6362\u4F1A\u8BDD / \u4F1A\u8BDD\u5386\u53F2 / \u6362\u6A21\u578B / \u601D\u8003\u5F3A\u5EA6 / \u8BA1\u5212\u6A21\u5F0F / \u6362\u9879\u76EE / \u5EFA\u7FA4 / \u6362\u56E2\u961F / \u6362\u5BF9\u8BDD\u3002",
     sessions: "\xB7 /sessions \u2014\u2014 \u5217\u51FA\u672C\u9879\u76EE\u7684\u4F1A\u8BDD\uFF1B\u52A0 /repo \u91CC\u7684\u9879\u76EE\u7F16\u53F7\uFF0C\u5217\u51FA\u672C\u673A\u5668\u4EBA\u5728\u522B\u7684\u9879\u76EE\u5F00\u8FC7\u7684",
     resume: "\xB7 /resume \u7F16\u53F7\u6216 id \u5F00\u5934 \u2014\u2014 \u5207\u4F1A\u8BDD\uFF1B\u522B\u7684\u9879\u76EE\u7684\u4F1A\u8BDD\u4F1A\u8FDE\u540C\u9879\u76EE\u4E00\u8D77\u5207",
     history: "\xB7 /history [\u6570\u91CF] \u2014\u2014 \u4EC5\u79C1\u804A\uFF1A\u67E5\u770B\u5F53\u524D\u4F1A\u8BDD\u6700\u8FD1\u7684\u6D88\u606F\uFF08\u9ED8\u8BA4 {{defaultCount}} \u6761\uFF0C\u6700\u591A {{maxCount}} \u6761\uFF09",
     model: "\xB7 /model \u2014\u2014 \u770B\u8FD9\u4E2A\u804A\u5929\u7528\u7684\u6A21\u578B\uFF1B/model \u7F16\u53F7 \u6216 /model \u5B9E\u4F8B/\u6A21\u578B \u5207\u6362",
+    reasoning: "\xB7 /reasoning \u2014\u2014 \u770B\u8FD9\u4E2A\u804A\u5929\u5F53\u524D\u6A21\u578B\u7684\u601D\u8003\u5F3A\u5EA6\uFF1B/reasoning \u53D6\u503C \u6216 /reasoning \u7F16\u53F7 \u5207\u6362\uFF1B/reasoning default \u6062\u590D\u6A21\u578B\u9ED8\u8BA4",
     preset: "\xB7 /preset \u2014\u2014 \u770B\u8FD9\u4E2A\u804A\u5929\u65B0\u4F1A\u8BDD\u7528\u7684\u9884\u8BBE\uFF1B/preset \u7F16\u53F7 \u6216 /preset \u540D\u5B57\u6216 id \u5207\u6362\uFF1B/preset none \u6E05\u6389\u804A\u5929\u9009\u62E9",
     plan: "\xB7 /plan \u2014\u2014 \u5F00\u5173\u8BA1\u5212\u6A21\u5F0F\uFF08agent \u5148\u51FA\u8BA1\u5212\u3001\u4E0D\u52A8\u624B\uFF09\uFF1B/plan on\u3001/plan off \u660E\u786E\u8BBE\u5B9A",
     panel: "\u53D1 /panel \u6253\u5F00\u53EF\u70B9\u7684\u547D\u4EE4\u9762\u677F\u3002",
@@ -147196,6 +147532,28 @@ var zh_CN_default = {
     group: "\xB7 /group \u7FA4\u540D \u2014\u2014 \u4EC5\u98DE\u4E66\uFF1A\u673A\u5668\u4EBA\u5EFA\u4E00\u4E2A\u7FA4\u5E76\u628A\u4F60\u62C9\u8FDB\u53BB\uFF08\u522B\u7684\u5E73\u53F0\u7684\u673A\u5668\u4EBA\u5EFA\u4E0D\u4E86\u7FA4\uFF09",
     syncNote: "\xB7 \u79C1\u804A\u9644\u7740\u4F1A\u8BDD\u540E\uFF0C\u7F51\u9875\u4E0A\u7684\u7528\u6237\u6587\u5B57\u4E0E\u7EC8\u6001\u52A9\u624B\u6587\u5B57\u4F1A\u540C\u6B65\u5230\u6B64\uFF08\u7FA4\u804A\u4E0D\u540C\u6B65\uFF09\u3002"
   },
+  reasoning: {
+    current: "\u5F53\u524D\u601D\u8003\u5F3A\u5EA6\uFF1A{{value}}",
+    currentUnlisted: "\u5F53\u524D\u601D\u8003\u5F3A\u5EA6\uFF1A{{value}}\uFF08\u5DF2\u4E0D\u5728\u53EF\u9009\u5217\u8868\uFF0C\u6362\u6389\u4E4B\u524D\u4ECD\u751F\u6548\uFF09",
+    resetNow: "\u601D\u8003\u5F3A\u5EA6\u5DF2\u6062\u590D\u4E3A\u6A21\u578B\u9ED8\u8BA4\uFF0C\u5F53\u524D\u4F1A\u8BDD\u4ECE\u4E0B\u4E00\u8F6E\u8D77\u4F7F\u7528\u5B83\uFF0C\u8FD9\u4E2A\u804A\u5929\u7684\u65B0\u4F1A\u8BDD\u4E5F\u4F7F\u7528\u5B83\u3002",
+    resetNext: "\u601D\u8003\u5F3A\u5EA6\u5DF2\u6062\u590D\u4E3A\u6A21\u578B\u9ED8\u8BA4\uFF0C\u8FD9\u4E2A\u804A\u5929\u7684\u65B0\u4F1A\u8BDD\u5C06\u4F7F\u7528\u5B83\u3002",
+    notFound: "\u300C{{query}}\u300D\u4E0D\u662F\u53EF\u9009\u7684\u601D\u8003\u5F3A\u5EA6\u3002",
+    numberOutOfRange: "\u7F16\u53F7 {{number}} \u8D85\u51FA\u8303\u56F4\uFF0C\u53EF\u9009 1\u2013{{max}}\u3002",
+    listTitle: "\u53EF\u9009\u601D\u8003\u5F3A\u5EA6\uFF08\u2605 \u5F53\u524D\uFF09\uFF1A",
+    listTitleDefault: "\u53EF\u9009\u601D\u8003\u5F3A\u5EA6\uFF08\u672A\u624B\u52A8\u9009\u62E9\uFF0C\u5F53\u524D\u7528\u6A21\u578B\u9ED8\u8BA4\uFF09\uFF1A",
+    more: "\u8FD8\u6709 {{count}} \u9879\u672A\u5217\u51FA\u3002",
+    usage: "\u7528\u6CD5\uFF1A/reasoning \u2014\u2014 \u770B\u8FD9\u4E2A\u804A\u5929\u5F53\u524D\u6A21\u578B\u7684\u601D\u8003\u5F3A\u5EA6\uFF1B/reasoning \u53D6\u503C \u6216 /reasoning \u7F16\u53F7 \u5207\u6362\uFF1B/reasoning default \u6062\u590D\u6A21\u578B\u9ED8\u8BA4\u3002",
+    choices: "\u53EF\u9009\uFF1A{{values}}\u3002",
+    unsupported: "\u5F53\u524D\u6A21\u578B\u4E0D\u652F\u6301\u5207\u6362\u601D\u8003\u5F3A\u5EA6\u3002",
+    readFailed: "\u6682\u65F6\u8BFB\u4E0D\u5230\u601D\u8003\u5F3A\u5EA6\u9009\u9879\uFF0C\u8BF7\u7A0D\u540E\u91CD\u8BD5\u3002",
+    setFailed: "\u5207\u6362\u601D\u8003\u5F3A\u5EA6\u5931\u8D25\uFF0C\u8BF7\u7A0D\u540E\u91CD\u8BD5\u3002",
+    setNow: "\u601D\u8003\u5F3A\u5EA6\u5DF2\u8BBE\u4E3A {{value}}\uFF0C\u5F53\u524D\u4F1A\u8BDD\u4ECE\u4E0B\u4E00\u8F6E\u8D77\u4F7F\u7528\u5B83\uFF0C\u8FD9\u4E2A\u804A\u5929\u7684\u65B0\u4F1A\u8BDD\u4E5F\u4F7F\u7528\u5B83\u3002",
+    setNext: "\u601D\u8003\u5F3A\u5EA6\u5DF2\u8BBE\u4E3A {{value}}\uFF0C\u8FD9\u4E2A\u804A\u5929\u7684\u65B0\u4F1A\u8BDD\u5C06\u4F7F\u7528\u5B83\u3002",
+    cleared: "\u65B0\u6A21\u578B\u4E0D\u652F\u6301\u539F\u6765\u7684\u601D\u8003\u5F3A\u5EA6\uFF08{{value}}\uFF09\uFF0C\u5DF2\u6E05\u9664\uFF0C\u6539\u7528\u6A21\u578B\u9ED8\u8BA4\u3002",
+    noModel: "\u8FD9\u4E2A\u804A\u5929\u8FD8\u6CA1\u6709\u6A21\u578B\uFF0C\u5148\u7528 /model \u9009\u4E00\u4E2A\u3002",
+    switchReadFailed: "\u6682\u65F6\u8BFB\u4E0D\u5230\u65B0\u6A21\u578B\u7684\u601D\u8003\u5F3A\u5EA6\u9009\u9879\uFF0C\u6A21\u578B\u6CA1\u6709\u5207\u6362\uFF0C\u8BF7\u7A0D\u540E\u91CD\u8BD5\u3002",
+    listTitlePlain: "\u53EF\u9009\u601D\u8003\u5F3A\u5EA6\uFF1A"
+  },
   history: {
     usage: "\u7528\u6CD5\uFF1A/history [\u6570\u91CF] \u2014\u2014 1\u2013{{maxCount}} \u6761\uFF08\u8D85\u51FA\u6309 {{maxCount}} \u6761\u7B97\uFF09\uFF0C\u9ED8\u8BA4 {{defaultCount}} \u6761\u3002",
     directOnly: "\u5728\u7FA4\u91CC\u4F1A\u628A\u4F1A\u8BDD\u5185\u5BB9\u53D1\u7ED9\u6240\u6709\u4EBA\u3002\u5728\u4E0E\u673A\u5668\u4EBA\u7684\u79C1\u804A\u91CC\u53D1 /history\u3002",
@@ -148554,47 +148912,17 @@ var makeLarkAdapter = (deps) => {
 };
 
 // src/modelPlan.ts
-import { threadOrNull as threadOrNull4 } from "@ycodium-ai/plugin-api";
-
-// src/botModel.ts
-var resolveDefaultModel = async (bot, chatModel, projectDefault, providers, knownInstances) => {
-  if (chatModel !== null) return chatModel;
-  if (bot.defaultModel !== void 0) {
-    const instances = knownInstances === void 0 ? await providers.list({}).catch(() => null) : knownInstances;
-    if (!isModelCatalogComplete(instances) || modelCatalogHas(modelChoices(instances ?? []), bot.defaultModel)) {
-      return bot.defaultModel;
-    }
-  }
-  return projectDefault;
-};
-
-// src/modelPlan.ts
-var makeModelPlan = (core, deps) => {
+var makeModelPlan = (core) => {
   const { host, t, commit, deliver } = core;
-  const { threadStillUsable } = deps;
+  const resolveCurrentModel = makeCurrentModelResolver(core);
   const currentModelOf = async (binding, instances) => {
-    const chat = chatOf(core.state, chatKeyOf(binding));
-    if (chat.threadId !== null) {
-      const thread = await threadOrNull4(host.threads, chat.threadId).catch(() => null);
-      if (thread !== null) {
-        return {
-          instanceId: thread.modelSelection.instanceId,
-          model: thread.modelSelection.model
-        };
-      }
-    }
-    if (chat.model !== null) return chat.model;
-    const project = binding.projectId === "" ? null : await host.projects.get({ projectId: binding.projectId }).catch(() => null);
-    return resolveDefaultModel(
-      binding,
-      chat.model,
-      project?.defaultModelSelection ?? null,
-      host.providers.instances,
-      instances
-    );
+    const { selection } = await resolveCurrentModel(binding, {
+      instances,
+      onReadError: () => null
+    });
+    return selection;
   };
   const handleModel = async (binding, arg) => {
-    const chatKey = chatKeyOf(binding);
     const instances = await host.providers.instances.list({}).catch(() => null);
     const choices = instances === null ? [] : modelChoices(instances);
     if (arg === "") {
@@ -148610,25 +148938,42 @@ var makeModelPlan = (core, deps) => {
       await deliver(binding, t("model.notFound", { query: arg }));
       return;
     }
-    const selection = { instanceId: picked.instanceId, model: picked.model };
-    commit("chat-updated", { chatKey, model: selection });
-    const threadId = chatOf(core.state, chatKey).threadId;
-    if (threadId !== null && await threadStillUsable(binding, threadId)) {
+    const { selection: current, threadId } = await resolveCurrentModel(binding, {
+      instances,
+      onReadError: () => null
+    });
+    const previous = selectedReasoning(current);
+    let carried;
+    if (previous !== void 0) {
       try {
-        await host.threads.update({
-          requestId: crypto.randomUUID(),
-          threadId,
-          modelSelection: selection
-        });
-      } catch (error) {
-        host.log.warn("chat could not switch the thread model", { error: describeError(error) });
-        await deliver(binding, t("model.setFailed"));
+        const catalog = await host.providers.models.list({ instanceId: picked.instanceId });
+        const model = catalog.models.find((entry) => entry.slug === picked.model);
+        const option2 = model === void 0 ? void 0 : reasoningOption(model);
+        if (option2 !== void 0 && option2.values.includes(previous.value))
+          carried = { id: option2.id, value: previous.value };
+      } catch {
+        host.log.warn("chat could not read reasoning options for a model switch");
+        await deliver(binding, t("reasoning.switchReadFailed"));
         return;
       }
-      await deliver(binding, t("model.setNow", { model: modelLabel(picked) }));
+    }
+    const selection = {
+      instanceId: picked.instanceId,
+      model: picked.model,
+      ...carried === void 0 ? {} : { options: [carried] }
+    };
+    const cleared = previous !== void 0 && carried === void 0 ? [t("reasoning.cleared", { value: previous.value })] : [];
+    if (!await saveModelSelection(core, binding, threadId, selection)) {
+      await deliver(binding, t("model.setFailed"));
       return;
     }
-    await deliver(binding, t("model.setNext", { model: modelLabel(picked) }));
+    await deliver(
+      binding,
+      [
+        t(threadId === null ? "model.setNext" : "model.setNow", { model: modelLabel(picked) }),
+        ...cleared
+      ].join("\n")
+    );
   };
   const handlePlan = async (binding, mode) => {
     if (mode === "invalid") {
@@ -148645,7 +148990,7 @@ var makeModelPlan = (core, deps) => {
     commit("chat-updated", { chatKey, plan: next });
     await deliver(binding, t(next ? "plan.on" : "plan.off"));
   };
-  return { currentModelOf, handleModel, handlePlan };
+  return { currentModelOf, resolveCurrentModel, handleModel, handlePlan };
 };
 
 // src/notifications.ts
@@ -151843,7 +152188,7 @@ var makeTurns = (core, deps) => {
   const threadStillUsable = async (binding, threadId) => {
     try {
       const thread = await threadOrNull8(host.threads, threadId, { archived: "include" });
-      return thread !== null && thread.archivedAt == null && thread.projectId === binding.projectId;
+      return threadUsable(binding, thread);
     } catch {
       return true;
     }
@@ -152108,7 +152453,6 @@ var makeTurns = (core, deps) => {
   const hasWaitingAttachments = (chatKey) => (waitingFor(chatKey, chatOf(core.state, chatKey).threadId)?.attachments.length ?? 0) > 0;
   return {
     stop: () => dispatchAbort.abort(),
-    threadStillUsable,
     ensureThread,
     startTurn,
     receiveAttachmentsOnly,
@@ -155597,7 +155941,7 @@ var makeWecomAppAdapter = (deps) => {
 // src/index.ts
 var definition = {
   id: "im",
-  apiVersion: 28,
+  apiVersion: 30,
   // 第二个参数只给插件自己的测试夹具：换掉文件 IO 与派发重试的等待。宿主只传 ctx。
   activate: async (ctx, testing = {}) => {
     const { dispatchSleep } = testing;
@@ -155742,7 +156086,7 @@ var definition = {
     });
     const projectRows = makeProjectRows(core);
     const sessions = makeSessions(core, projectRows);
-    const modelPlan = makeModelPlan(core, turns);
+    const modelPlan = makeModelPlan(core);
     const chatQueue = makeChatQueue(core);
     const panelWiring = makePanelWiring(core, {
       enqueueForChat: chatQueue.enqueueForChat,

package/locales/en.json

@@ -340,19 +340,25 @@
     "project": "Project: {{title}}",
     "noProject": "No project picked yet. Send /repo to pick one."
   },
+  "version": {
+    "current": "Ycodium Chat bridge v{{version}}",
+    "usage": "Usage: /version (no arguments)."
+  },
   "help": {
     "title": "Commands:",
     "continue": "· /continue — resume sending after /stop or a failure paused it; the waiting messages go out",
     "stop": "· /stop — stop the turn that is running; the session stays",
     "status": "· /status — what is going on right now",
+    "version": "· /version — the chat bridge's version (no arguments)",
     "new": "· /new — forget the current session; the next message starts a new one",
     "team": "· /team — give this project's team a goal with each message; use /repo to pick the project first",
     "thread": "· /thread — go back to a normal session",
-    "aliases": "Chinese shortcuts (停 / 继续发送 / 状态 / 新对话 / 会话列表 / 切换会话 / 会话历史 / 换模型 / 计划模式 / 换项目 / 建群 / 换团队 / 换对话) work when the reply language is Chinese.",
+    "aliases": "Chinese shortcuts (停 / 继续发送 / 状态 / 新对话 / 会话列表 / 切换会话 / 会话历史 / 换模型 / 思考强度 / 计划模式 / 换项目 / 建群 / 换团队 / 换对话) work when the reply language is Chinese.",
     "sessions": "· /sessions — list this project's sessions; add a project number from /repo to list the ones this bot opened in another project",
     "resume": "· /resume number-or-id-prefix — switch sessions; one from another project switches the project too",
     "history": "· /history [count] — preview recent messages in the current session, in private chats only (default {{defaultCount}}, maximum {{maxCount}})",
     "model": "· /model — see the model this chat uses; /model number or /model instance/model switches it",
+    "reasoning": "· /reasoning — see the reasoning effort for this chat's current model; /reasoning value or /reasoning number switches it; /reasoning default resets it to the model default",
     "preset": "· /preset — see the preset this chat uses for new sessions; /preset number or /preset name-or-id switches it; /preset none clears the chat choice",
     "plan": "· /plan — toggle plan mode (the agent proposes a plan instead of acting); /plan on, /plan off set it",
     "panel": "Send /panel to open the command panel with a button for every command.",
@@ -360,6 +366,28 @@
     "group": "· /group name — Lark only: the bot creates a group and adds you (bots on other platforms cannot)",
     "syncNote": "· After a private chat attaches a session, web user text and final assistant text sync here (groups never sync)."
   },
+  "reasoning": {
+    "current": "Current reasoning effort: {{value}}",
+    "currentUnlisted": "Current reasoning effort: {{value}} (no longer in the list; it stays in effect until you change it)",
+    "resetNow": "Reasoning effort reset to the model default. The current session uses it from the next turn; new sessions in this chat use it too.",
+    "resetNext": "Reasoning effort reset to the model default. New sessions in this chat will use it.",
+    "notFound": "\"{{query}}\" is not an available reasoning effort.",
+    "numberOutOfRange": "Number {{number}} is out of range; choose 1–{{max}}.",
+    "listTitle": "Available reasoning efforts (★ current):",
+    "listTitleDefault": "Available reasoning efforts (none chosen; the model default is used):",
+    "more": "{{count}} more not shown.",
+    "usage": "Usage: /reasoning — see the reasoning effort for this chat's current model; /reasoning value or /reasoning number switches it; /reasoning default resets it to the model default.",
+    "choices": "Choices: {{values}}.",
+    "unsupported": "The current model does not support switching the reasoning effort.",
+    "readFailed": "Could not read the reasoning options. Try again in a moment.",
+    "setFailed": "Switching the reasoning effort failed. Try again in a moment.",
+    "setNow": "Reasoning effort set to {{value}}. The current session uses it from the next turn; new sessions in this chat use it too.",
+    "setNext": "Reasoning effort set to {{value}}. New sessions in this chat will use it.",
+    "cleared": "The new model does not support the previous reasoning effort ({{value}}). It was cleared; the model default will be used.",
+    "noModel": "This chat has no model yet. Pick one with /model first.",
+    "switchReadFailed": "Could not read the new model's reasoning efforts, so the model was not switched. Try again in a moment.",
+    "listTitlePlain": "Available reasoning efforts:"
+  },
   "history": {
     "usage": "Usage: /history [count] — 1 to {{maxCount}} (larger numbers count as {{maxCount}}), default {{defaultCount}}.",
     "directOnly": "In a group this would show the session to everyone. Send /history in a private chat with the bot.",

package/locales/zh-CN.json

@@ -340,19 +340,25 @@
     "project": "项目：{{title}}",
     "noProject": "还没选项目。发 /repo 选一个。"
   },
+  "version": {
+    "current": "Ycodium 聊天桥 v{{version}}",
+    "usage": "用法：/version（不带参数）。"
+  },
   "help": {
     "title": "能用的命令：",
     "continue": "· /continue —— /stop 或出错暂停了发送之后，用它恢复，排着的消息继续发出",
     "stop": "· /stop —— 停掉正在跑的这一轮，会话留着",
     "status": "· /status —— 现在在干什么",
+    "version": "· /version —— 聊天桥的版本（不带参数）",
     "new": "· /new —— 忘掉当前会话，下一句开新的",
     "team": "· /team —— 每条消息都给这个项目的团队开一个目标；先用 /repo 选项目",
     "thread": "· /thread —— 切回普通会话",
-    "aliases": "中文也行：停 / 继续发送 / 状态 / 新对话 / 会话列表 / 切换会话 / 会话历史 / 换模型 / 计划模式 / 换项目 / 建群 / 换团队 / 换对话。",
+    "aliases": "中文也行：停 / 继续发送 / 状态 / 新对话 / 会话列表 / 切换会话 / 会话历史 / 换模型 / 思考强度 / 计划模式 / 换项目 / 建群 / 换团队 / 换对话。",
     "sessions": "· /sessions —— 列出本项目的会话；加 /repo 里的项目编号，列出本机器人在别的项目开过的",
     "resume": "· /resume 编号或 id 开头 —— 切会话；别的项目的会话会连同项目一起切",
     "history": "· /history [数量] —— 仅私聊：查看当前会话最近的消息（默认 {{defaultCount}} 条，最多 {{maxCount}} 条）",
     "model": "· /model —— 看这个聊天用的模型；/model 编号 或 /model 实例/模型 切换",
+    "reasoning": "· /reasoning —— 看这个聊天当前模型的思考强度；/reasoning 取值 或 /reasoning 编号 切换；/reasoning default 恢复模型默认",
     "preset": "· /preset —— 看这个聊天新会话用的预设；/preset 编号 或 /preset 名字或 id 切换；/preset none 清掉聊天选择",
     "plan": "· /plan —— 开关计划模式（agent 先出计划、不动手）；/plan on、/plan off 明确设定",
     "panel": "发 /panel 打开可点的命令面板。",
@@ -360,6 +366,28 @@
     "group": "· /group 群名 —— 仅飞书：机器人建一个群并把你拉进去（别的平台的机器人建不了群）",
     "syncNote": "· 私聊附着会话后，网页上的用户文字与终态助手文字会同步到此（群聊不同步）。"
   },
+  "reasoning": {
+    "current": "当前思考强度：{{value}}",
+    "currentUnlisted": "当前思考强度：{{value}}（已不在可选列表，换掉之前仍生效）",
+    "resetNow": "思考强度已恢复为模型默认，当前会话从下一轮起使用它，这个聊天的新会话也使用它。",
+    "resetNext": "思考强度已恢复为模型默认，这个聊天的新会话将使用它。",
+    "notFound": "「{{query}}」不是可选的思考强度。",
+    "numberOutOfRange": "编号 {{number}} 超出范围，可选 1–{{max}}。",
+    "listTitle": "可选思考强度（★ 当前）：",
+    "listTitleDefault": "可选思考强度（未手动选择，当前用模型默认）：",
+    "more": "还有 {{count}} 项未列出。",
+    "usage": "用法：/reasoning —— 看这个聊天当前模型的思考强度；/reasoning 取值 或 /reasoning 编号 切换；/reasoning default 恢复模型默认。",
+    "choices": "可选：{{values}}。",
+    "unsupported": "当前模型不支持切换思考强度。",
+    "readFailed": "暂时读不到思考强度选项，请稍后重试。",
+    "setFailed": "切换思考强度失败，请稍后重试。",
+    "setNow": "思考强度已设为 {{value}}，当前会话从下一轮起使用它，这个聊天的新会话也使用它。",
+    "setNext": "思考强度已设为 {{value}}，这个聊天的新会话将使用它。",
+    "cleared": "新模型不支持原来的思考强度（{{value}}），已清除，改用模型默认。",
+    "noModel": "这个聊天还没有模型，先用 /model 选一个。",
+    "switchReadFailed": "暂时读不到新模型的思考强度选项，模型没有切换，请稍后重试。",
+    "listTitlePlain": "可选思考强度："
+  },
   "history": {
     "usage": "用法：/history [数量] —— 1–{{maxCount}} 条（超出按 {{maxCount}} 条算），默认 {{defaultCount}} 条。",
     "directOnly": "在群里会把会话内容发给所有人。在与机器人的私聊里发 /history。",

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ycodium-ai/plugin-im",
-  "version": "0.2.3250",
+  "version": "0.2.3254",
   "description": "Talk to your agents from Telegram, WeCom, Lark, or WeChat (experimental).",
   "keywords": [
     "im",
@@ -33,7 +33,7 @@
     "react": "19.2.6",
     "vite-plus": "0.2.2",
     "ycodium-react-test-harness": "0.0.0",
-    "@ycodium-ai/plugin-api": "0.2.3250"
+    "@ycodium-ai/plugin-api": "0.2.3254"
   },
   "ycodium": {
     "plugin": "./dist/index.js",