jinzd-ai-cli 0.4.67 → 0.4.68

package/dist/{chunk-F44OKMB2.js → chunk-3LCVJ4AF.js}

@@ -8,7 +8,7 @@ import { platform } from "os";
 import chalk from "chalk";
 
 // src/core/constants.ts
-var VERSION = "0.4.67";
+var VERSION = "0.4.68";
 var APP_NAME = "ai-cli";
 var CONFIG_DIR_NAME = ".aicli";
 var CONFIG_FILE_NAME = "config.json";

package/dist/{chunk-XSIVGDCN.js → chunk-G5AISHJE.js}

@@ -8,7 +8,7 @@ import {
   RateLimitError,
   schemaToJsonSchema,
   truncateForPersist
-} from "./chunk-BVLQ3FRA.js";
+} from "./chunk-Q5QSCO5D.js";
 import {
   APP_NAME,
   CONFIG_DIR_NAME,
@@ -21,7 +21,7 @@ import {
   MCP_TOOL_PREFIX,
   PLUGINS_DIR_NAME,
   VERSION
-} from "./chunk-F44OKMB2.js";
+} from "./chunk-3LCVJ4AF.js";
 
 // src/config/config-manager.ts
 import { readFileSync, writeFileSync, existsSync, mkdirSync } from "fs";
@@ -191,7 +191,27 @@ var ConfigSchema = z.object({
   // 插件以完整 Node.js 权限在主进程中执行（可读写文件、访问网络、执行命令），
   // 必须确认插件来源可信后，再设为 true 启用。
   // 可通过 /config 命令或直接编辑 ~/.aicli/config.json 开启。
-  allowPlugins: z.boolean().default(false)
+  allowPlugins: z.boolean().default(false),
+  // 智能模型路由（v0.4.68+）
+  // 按用户每轮输入的内容/标签/长度动态选择模型，在同一 provider 内切换，
+  // 例：短问题走 haiku（省钱），planning 走 opus（质量）。
+  // enabled=false 时永远返回当前模型。rules 按顺序匹配，首个命中的规则生效。
+  // 每个 rule 的 match 必须至少有一个条件（tag/contains/maxLength/minLength）。
+  // 详见 src/core/model-router.ts。
+  routing: z.object({
+    enabled: z.boolean().default(false),
+    rules: z.array(z.object({
+      match: z.object({
+        contains: z.array(z.string()).optional(),
+        maxLength: z.number().int().positive().optional(),
+        minLength: z.number().int().positive().optional(),
+        tag: z.string().optional()
+      }),
+      model: z.string(),
+      name: z.string().optional()
+    })).default([]),
+    fallback: z.string().optional()
+  }).default({ enabled: false, rules: [] })
 });
 
 // src/config/config-manager.ts

package/dist/{chunk-BVLQ3FRA.js → chunk-Q5QSCO5D.js}

@@ -10,7 +10,7 @@ import {
   SUBAGENT_DEFAULT_MAX_ROUNDS,
   SUBAGENT_MAX_ROUNDS_LIMIT,
   runTestsTool
-} from "./chunk-F44OKMB2.js";
+} from "./chunk-3LCVJ4AF.js";
 
 // src/tools/builtin/bash.ts
 import { execSync } from "child_process";

package/dist/{chunk-X7NVAEFI.js → chunk-VO5IZN2C.js}

@@ -6,7 +6,7 @@ import { platform } from "os";
 import chalk from "chalk";
 
 // src/core/constants.ts
-var VERSION = "0.4.67";
+var VERSION = "0.4.68";
 var APP_NAME = "ai-cli";
 var CONFIG_DIR_NAME = ".aicli";
 var CONFIG_FILE_NAME = "config.json";

package/dist/{hub-RGK325NQ.js → hub-4VPTOMBP.js}

@@ -385,7 +385,7 @@ ${content}`);
   }
 }
 async function runTaskMode(config, providers, configManager, topic) {
-  const { TaskOrchestrator } = await import("./task-orchestrator-4PVBMQJ7.js");
+  const { TaskOrchestrator } = await import("./task-orchestrator-WDRXASIC.js");
   const orchestrator = new TaskOrchestrator(config, providers, configManager);
   let interrupted = false;
   const onSigint = () => {

package/dist/index.js

@@ -31,7 +31,7 @@ import {
   saveDevState,
   sessionHasMeaningfulContent,
   setupProxy
-} from "./chunk-XSIVGDCN.js";
+} from "./chunk-G5AISHJE.js";
 import {
   ToolExecutor,
   ToolRegistry,
@@ -47,7 +47,7 @@ import {
   spawnAgentContext,
   theme,
   undoStack
-} from "./chunk-BVLQ3FRA.js";
+} from "./chunk-Q5QSCO5D.js";
 import {
   fileCheckpoints
 } from "./chunk-4BKXL7SM.js";
@@ -72,7 +72,7 @@ import {
   SKILLS_DIR_NAME,
   VERSION,
   buildUserIdentityPrompt
-} from "./chunk-F44OKMB2.js";
+} from "./chunk-3LCVJ4AF.js";
 
 // src/index.ts
 import { program } from "commander";
@@ -1075,6 +1075,79 @@ function createDefaultCommands() {
         );
       }
     },
+    {
+      name: "route",
+      description: "Smart model routing \u2014 enable/disable or inspect routing rules",
+      usage: "/route [on|off|show|test <message>]",
+      async execute(args, ctx) {
+        const sub = (args[0] ?? "show").toLowerCase();
+        const routing = ctx.config.get("routing");
+        if (sub === "on" || sub === "enable") {
+          ctx.config.setByPath("routing.enabled", "true");
+          ctx.renderer.printSuccess("Smart model routing enabled.");
+          if (!routing || routing.rules.length === 0) {
+            ctx.renderer.printInfo(
+              'No rules configured yet. Add rules under `routing.rules` in ~/.aicli/config.json.\nExample: { match: { tag: "fast" }, model: "claude-haiku-4-5" }'
+            );
+          }
+          return;
+        }
+        if (sub === "off" || sub === "disable") {
+          ctx.config.setByPath("routing.enabled", "false");
+          ctx.renderer.printSuccess("Smart model routing disabled.");
+          return;
+        }
+        if (sub === "test") {
+          const msg = args.slice(1).join(" ").trim();
+          if (!msg) {
+            console.log(theme.warning("  Usage: /route test <message>"));
+            return;
+          }
+          const decision = ctx.computeRoutingDecision(msg);
+          const marker = decision.overridden ? theme.accent("\u2192 ROUTED") : theme.dim("(unchanged)");
+          console.log();
+          console.log(`  Input:    ${theme.dim(msg)}`);
+          console.log(`  Current:  ${theme.info(ctx.getCurrentModel())}`);
+          console.log(`  Decision: ${theme.info(decision.model)}  ${marker}`);
+          console.log(`  Reason:   ${theme.dim(decision.reason)}`);
+          if (typeof decision.ruleIdx === "number") {
+            console.log(`  Rule:     #${decision.ruleIdx}`);
+          }
+          console.log();
+          return;
+        }
+        if (sub === "show" || sub === "status") {
+          console.log();
+          console.log(`  ${theme.heading("Smart Model Routing")}`);
+          console.log(`  Status:   ${routing?.enabled ? theme.success("enabled") : theme.dim("disabled")}`);
+          console.log(`  Provider: ${theme.info(ctx.getCurrentProvider())}`);
+          console.log(`  Current:  ${theme.info(ctx.getCurrentModel())}`);
+          if (routing?.fallback) {
+            console.log(`  Fallback: ${theme.info(routing.fallback)}`);
+          }
+          console.log();
+          if (!routing || routing.rules.length === 0) {
+            console.log(`  ${theme.dim("(no rules configured \u2014 edit ~/.aicli/config.json `routing.rules`)")}`);
+          } else {
+            console.log(`  ${theme.heading("Rules")} ${theme.dim(`(evaluated top-to-bottom)`)}:`);
+            routing.rules.forEach((r, i) => {
+              const parts = [];
+              if (r.match.tag) parts.push(`tag=#${r.match.tag}`);
+              if (r.match.contains && r.match.contains.length > 0) parts.push(`contains=[${r.match.contains.slice(0, 3).join(", ")}${r.match.contains.length > 3 ? ", \u2026" : ""}]`);
+              if (typeof r.match.maxLength === "number") parts.push(`maxLen=${r.match.maxLength}`);
+              if (typeof r.match.minLength === "number") parts.push(`minLen=${r.match.minLength}`);
+              const cond = parts.length > 0 ? parts.join(" & ") : theme.warning("(empty \u2014 never matches)");
+              console.log(`    ${theme.dim(`#${i}`)} ${r.name ? theme.accent(r.name) + " " : ""}${cond} ${theme.dim("\u2192")} ${theme.info(r.model)}`);
+            });
+          }
+          console.log();
+          console.log(`  ${theme.dim("Commands: /route on | off | test <msg> | show")}`);
+          console.log();
+          return;
+        }
+        console.log(theme.warning(`  Unknown subcommand: ${sub}. Usage: /route [on|off|show|test <message>]`));
+      }
+    },
     {
       name: "clear",
       description: "Clear conversation history",
@@ -1124,18 +1197,18 @@ function createDefaultCommands() {
             return;
           }
           const sessions = ctx.sessions.listSessions();
-          const matches = sessions.filter((s) => s.id.startsWith(id));
-          if (matches.length === 0) {
+          const matches2 = sessions.filter((s) => s.id.startsWith(id));
+          if (matches2.length === 0) {
             ctx.renderer.renderError(`Session '${id}' not found.`);
             return;
           }
-          if (matches.length > 1) {
-            console.log(theme.warning(`  \u26A0 Ambiguous prefix '${id}' matches ${matches.length} sessions \u2014 loading most recent:`));
-            for (const m of matches.slice(0, 5)) {
+          if (matches2.length > 1) {
+            console.log(theme.warning(`  \u26A0 Ambiguous prefix '${id}' matches ${matches2.length} sessions \u2014 loading most recent:`));
+            for (const m of matches2.slice(0, 5)) {
               console.log(theme.dim(`    ${m.id.slice(0, 12)}  ${m.title ?? "(untitled)"}`));
             }
           }
-          const match = matches[0];
+          const match = matches2[0];
           ctx.sessions.loadSession(match.id);
           ctx.setProvider(match.provider, match.model);
           ctx.resetSessionTokenUsage();
@@ -1373,13 +1446,13 @@ ${text}
   ${theme.heading(`Found ${results.length} session(s) containing "${query}"`)}
 `);
         for (const r of results) {
-          const { sessionMeta, matches } = r;
+          const { sessionMeta, matches: matches2 } = r;
           const dateStr = sessionMeta.updated.toLocaleDateString();
           console.log(
             `  ${theme.accent(sessionMeta.id.slice(0, 8))}` + theme.dim(`  [${dateStr}]  ${sessionMeta.provider} / ${sessionMeta.model}`) + (sessionMeta.title ? `
   ${theme.dim("  " + sessionMeta.title)}` : "")
           );
-          for (const m of matches) {
+          for (const m of matches2) {
             const icon = m.role === "user" ? "\u{1F464}" : "\u{1F916}";
             console.log(`    ${icon} ${theme.warning(m.snippet)}`);
           }
@@ -2194,7 +2267,7 @@ ${hint}` : "")
       usage: "/test [command|filter]",
       async execute(args, ctx) {
         try {
-          const { executeTests } = await import("./run-tests-NUF7CNM4.js");
+          const { executeTests } = await import("./run-tests-WD53PYVA.js");
           const argStr = args.join(" ").trim();
           let testArgs = {};
           if (argStr) {
@@ -3394,6 +3467,74 @@ var CostTracker = class {
   }
 };
 
+// src/core/model-router.ts
+var TAG_REGEX = /(?:^|\s)#([a-zA-Z][\w-]{0,31})\b/g;
+function extractTags(message) {
+  const tags = /* @__PURE__ */ new Set();
+  let m;
+  TAG_REGEX.lastIndex = 0;
+  while ((m = TAG_REGEX.exec(message)) !== null) {
+    tags.add(m[1].toLowerCase());
+  }
+  return tags;
+}
+function matches(message, matcher) {
+  const trimmed = message.trim();
+  const lower = trimmed.toLowerCase();
+  if (matcher.tag) {
+    const tags = extractTags(trimmed);
+    if (!tags.has(matcher.tag.toLowerCase())) return false;
+  }
+  if (matcher.contains && matcher.contains.length > 0) {
+    const hit = matcher.contains.some((kw) => lower.includes(kw.toLowerCase()));
+    if (!hit) return false;
+  }
+  if (typeof matcher.maxLength === "number") {
+    if (trimmed.length > matcher.maxLength) return false;
+  }
+  if (typeof matcher.minLength === "number") {
+    if (trimmed.length < matcher.minLength) return false;
+  }
+  const hasAnyCondition = !!matcher.tag || matcher.contains && matcher.contains.length > 0 || typeof matcher.maxLength === "number" || typeof matcher.minLength === "number";
+  return !!hasAnyCondition;
+}
+function pickModel(message, currentModel, config, availableModels = []) {
+  if (!config.enabled || config.rules.length === 0) {
+    return { model: currentModel, reason: "routing disabled", overridden: false };
+  }
+  const isAvailable = (m) => availableModels.length === 0 || availableModels.includes(m);
+  for (let i = 0; i < config.rules.length; i++) {
+    const rule = config.rules[i];
+    if (!matches(message, rule.match)) continue;
+    if (!isAvailable(rule.model)) continue;
+    if (rule.model === currentModel) {
+      return {
+        model: currentModel,
+        reason: `rule "${rule.name ?? `#${i}`}" matched (same as current)`,
+        overridden: false,
+        ruleIdx: i
+      };
+    }
+    return {
+      model: rule.model,
+      reason: `rule "${rule.name ?? `#${i}`}" matched`,
+      overridden: true,
+      ruleIdx: i
+    };
+  }
+  if (config.fallback && config.fallback !== currentModel && isAvailable(config.fallback)) {
+    return {
+      model: config.fallback,
+      reason: "fallback",
+      overridden: true
+    };
+  }
+  return { model: currentModel, reason: "no rule matched", overridden: false };
+}
+function stripRoutingTags(message) {
+  return message.replace(/(?:^|\s)#(fast|deep|default)\b/gi, " ").replace(/\s{2,}/g, " ").trim();
+}
+
 // src/repl/notify.ts
 import { spawn } from "child_process";
 import { platform as platform2 } from "os";
@@ -3570,13 +3711,15 @@ var Repl = class {
   contextLayers = [];
   /** 本次会话累计 token 用量 */
   sessionTokenUsage = { inputTokens: 0, outputTokens: 0, cacheCreationTokens: 0, cacheReadTokens: 0 };
-  /** Fold a single-request TokenUsage (with optional cache fields) into sessionTokenUsage + cost tracker. */
-  addSessionUsage(u) {
+  /** Fold a single-request TokenUsage (with optional cache fields) into sessionTokenUsage + cost tracker.
+   *  modelOverride lets the smart router attribute cost to the actually-used model
+   *  when it differs from the UI-selected currentModel. */
+  addSessionUsage(u, modelOverride) {
     this.sessionTokenUsage.inputTokens += u.inputTokens;
     this.sessionTokenUsage.outputTokens += u.outputTokens;
     this.sessionTokenUsage.cacheCreationTokens += u.cacheCreationTokens ?? 0;
     this.sessionTokenUsage.cacheReadTokens += u.cacheReadTokens ?? 0;
-    this.costTracker.addCost(this.currentProvider, this.currentModel, u);
+    this.costTracker.addCost(this.currentProvider, modelOverride ?? this.currentModel, u);
   }
   /** 启动时检测到的 Git 分支（无 git 仓库时为 null） */
   gitBranch = null;
@@ -4468,7 +4611,16 @@ Session '${this.resumeSessionId}' not found.
 `));
       }
     }
-    const messageContent = parts.length > 0 ? parts.length === 1 && parts[0].type === "text" ? parts[0].text : parts : userInput;
+    const routingDecision = this.computeRoutingDecision(userInput);
+    const cleanInput = stripRoutingTags(userInput);
+    let effectiveParts = parts;
+    if (cleanInput !== userInput && parts.length > 0 && parts[0].type === "text") {
+      effectiveParts = [
+        { type: "text", text: stripRoutingTags(parts[0].text ?? "") },
+        ...parts.slice(1)
+      ];
+    }
+    const messageContent = effectiveParts.length > 0 ? effectiveParts.length === 1 && effectiveParts[0].type === "text" ? effectiveParts[0].text : effectiveParts : cleanInput;
     if (hasImage) {
       const visionHint = this.getVisionModelHint();
       if (visionHint) {
@@ -4489,6 +4641,12 @@ Session '${this.resumeSessionId}' not found.
       timestamp: /* @__PURE__ */ new Date()
     });
     this.events.emit("message.before", { input: userInput });
+    if (routingDecision.overridden) {
+      process.stdout.write(
+        theme.dim(`  \u2192 Routed to ${routingDecision.model} (${routingDecision.reason})
+`)
+      );
+    }
     const t0 = Date.now();
     try {
       const provider = this.providers.get(this.currentProvider);
@@ -4496,10 +4654,11 @@ Session '${this.resumeSessionId}' not found.
       if (supportsTools) {
         await this.handleChatWithTools(
           provider,
-          session.messages
+          session.messages,
+          routingDecision.model
         );
       } else {
-        await this.handleChatSimple(provider, session.messages);
+        await this.handleChatSimple(provider, session.messages, routingDecision.model);
       }
       if (this.config.get("session").autoSave) {
         if (autoTrimSessionIfNeeded(session)) {
@@ -4577,9 +4736,10 @@ Session '${this.resumeSessionId}' not found.
    */
   /** 运行时 thinking 模式覆盖：null=使用配置值，true/false=运行时覆盖 */
   runtimeThinking = null;
-  getModelParams() {
+  getModelParams(modelOverride) {
     const allParams = this.config.get("modelParams");
-    const params = allParams[this.currentModel] ?? {};
+    const modelId = modelOverride ?? this.currentModel;
+    const params = allParams[modelId] ?? {};
     return {
       ...params,
       maxTokens: params.maxTokens ?? DEFAULT_MAX_TOKENS,
@@ -4587,6 +4747,25 @@ Session '${this.resumeSessionId}' not found.
       thinkingBudget: params.thinkingBudget
     };
   }
+  /**
+   * Compute smart-routing decision for this user turn.
+   * Only considers models available for the current provider (rule skipped otherwise).
+   * When routing is disabled or no rule matches, returns the current model unchanged.
+   */
+  computeRoutingDecision(userInput) {
+    const routingConfig = this.config.get("routing");
+    if (!routingConfig || !routingConfig.enabled) {
+      return { model: this.currentModel, reason: "routing disabled", overridden: false };
+    }
+    let availableModels = [];
+    try {
+      const provider = this.providers.get(this.currentProvider);
+      availableModels = provider.info.models.map((m) => m.id);
+    } catch {
+      availableModels = [];
+    }
+    return pickModel(userInput, this.currentModel, routingConfig, availableModels);
+  }
   // ─── Context 自动管理 ───────────────────────────────────────────────────
   /**
    * 估算文本的 token 数。
@@ -4639,12 +4818,15 @@ Session '${this.resumeSessionId}' not found.
     return total;
   }
   /**
-   * 获取当前模型的 context window 大小。
+   * 获取指定模型的 context window 大小（默认当前模型）。
+   * 智能路由可能在 handleChatWithTools 内把 effectiveModel 暂时切到别的模型，
+   * 故此处接受可选的 modelOverride 以保持计算一致性。
    */
-  getContextWindowSize() {
+  getContextWindowSize(modelOverride) {
     try {
       const provider = this.providers.get(this.currentProvider);
-      const modelInfo = provider.info.models.find((m) => m.id === this.currentModel);
+      const modelId = modelOverride ?? this.currentModel;
+      const modelInfo = provider.info.models.find((m) => m.id === modelId);
       return modelInfo?.contextWindow ?? 0;
     } catch {
       return 0;
@@ -4881,16 +5063,17 @@ Session '${this.resumeSessionId}' not found.
       }
     });
   }
-  async handleChatSimple(provider, messages) {
+  async handleChatSimple(provider, messages, modelOverride) {
     const session = this.sessions.current;
     const useStreaming = this.config.get("ui").streaming;
-    const modelParams = this.getModelParams();
+    const effectiveModel = modelOverride ?? this.currentModel;
+    const modelParams = this.getModelParams(effectiveModel);
     if (useStreaming) {
       const ac = this.setupStreamInterrupt();
       try {
         const stream = provider.chatStream({
           messages,
-          model: this.currentModel,
+          model: effectiveModel,
           systemPrompt: this.buildCurrentSystemPrompt(),
           stream: true,
           temperature: modelParams.temperature,
@@ -4910,7 +5093,7 @@ Session '${this.resumeSessionId}' not found.
         session.addMessage({ role: "assistant", content, timestamp: /* @__PURE__ */ new Date() });
         this.events.emit("message.after", { content });
         if (usage) {
-          this.addSessionUsage(usage);
+          this.addSessionUsage(usage, effectiveModel);
           session.addTokenUsage(usage);
           if (showTokens && !tokensShown) {
             this.renderer.renderUsage(usage, this.sessionTokenUsage);
@@ -4924,7 +5107,7 @@ Session '${this.resumeSessionId}' not found.
       try {
         const response = await provider.chat({
           messages,
-          model: this.currentModel,
+          model: effectiveModel,
           systemPrompt: this.buildCurrentSystemPrompt(),
           stream: false,
           temperature: modelParams.temperature,
@@ -4939,7 +5122,7 @@ Session '${this.resumeSessionId}' not found.
         session.addMessage({ role: "assistant", content: response.content, timestamp: /* @__PURE__ */ new Date() });
         this.events.emit("message.after", { content: response.content });
         if (response.usage) {
-          this.addSessionUsage(response.usage);
+          this.addSessionUsage(response.usage, effectiveModel);
           session.addTokenUsage(response.usage);
           if (this.shouldShowTokens()) {
             this.renderer.renderUsage(response.usage, this.sessionTokenUsage);
@@ -5052,8 +5235,9 @@ Session '${this.resumeSessionId}' not found.
       rawContent
     };
   }
-  async handleChatWithTools(provider, messages) {
+  async handleChatWithTools(provider, messages, modelOverride) {
     const session = this.sessions.current;
+    const effectiveModel = modelOverride ?? this.currentModel;
     let toolDefs;
     let mcpBudgetNote = null;
     const usedMcpToolNames = /* @__PURE__ */ new Set();
@@ -5064,7 +5248,7 @@ Session '${this.resumeSessionId}' not found.
       if (skillFilter) {
         toolDefs = this.toolRegistry.getDefinitions().filter((t) => skillFilter.has(t.name));
       } else {
-        const contextWindow = this.getContextWindowSize();
+        const contextWindow = this.getContextWindowSize(effectiveModel);
         if (contextWindow > 0) {
           const toolBudget = Math.floor(contextWindow * 0.2);
           const { definitions, trimmedCount, systemNote } = this.toolRegistry.getDefinitionsWithBudget(toolBudget, usedMcpToolNames);
@@ -5114,7 +5298,7 @@ You have a maximum of ${maxToolRounds} tool call rounds for this task. Plan effi
     const systemPrompt = baseSystemPrompt + roundBudgetHint + (mcpBudgetNote ? `
 
 ${mcpBudgetNote}` : "");
-    const modelParams = this.getModelParams();
+    const modelParams = this.getModelParams(effectiveModel);
     const useStreaming = this.config.get("ui").streaming;
     const spinner = this.renderer.showSpinner("Thinking...");
     const roundUsage = { inputTokens: 0, outputTokens: 0, cacheCreationTokens: 0, cacheReadTokens: 0 };
@@ -5204,7 +5388,7 @@ ${mcpBudgetNote}` : "");
               )
             );
             if (roundUsage.inputTokens > 0 || roundUsage.outputTokens > 0) {
-              this.addSessionUsage(roundUsage);
+              this.addSessionUsage(roundUsage, effectiveModel);
               session.addTokenUsage(roundUsage);
               if (this.shouldShowTokens()) {
                 this.renderer.renderUsage(roundUsage, this.sessionTokenUsage);
@@ -5232,7 +5416,7 @@ ${mcpBudgetNote}` : "");
         let alreadyRendered = false;
         const chatRequest = {
           messages: apiMessages,
-          model: this.currentModel,
+          model: effectiveModel,
           systemPrompt,
           stream: false,
           temperature: modelParams.temperature,
@@ -5334,7 +5518,7 @@ ${mcpBudgetNote}` : "");
               )
             );
             if (roundUsage.inputTokens > 0 || roundUsage.outputTokens > 0) {
-              this.addSessionUsage(roundUsage);
+              this.addSessionUsage(roundUsage, effectiveModel);
               session.addTokenUsage(roundUsage);
               if (this.shouldShowTokens()) {
                 this.renderer.renderUsage(roundUsage, this.sessionTokenUsage);
@@ -5369,7 +5553,7 @@ ${mcpBudgetNote}` : "");
           });
           this.events.emit("message.after", { content: finalContent });
           if (roundUsage.inputTokens > 0 || roundUsage.outputTokens > 0) {
-            this.addSessionUsage(roundUsage);
+            this.addSessionUsage(roundUsage, effectiveModel);
             session.addTokenUsage(roundUsage);
             if (this.shouldShowTokens()) {
               this.renderer.renderUsage(roundUsage, this.sessionTokenUsage);
@@ -5387,7 +5571,7 @@ ${mcpBudgetNote}` : "");
             try {
               const genStream = provider.chatStream({
                 messages: apiMessages,
-                model: this.currentModel,
+                model: effectiveModel,
                 systemPrompt,
                 stream: true,
                 temperature: modelParams.temperature,
@@ -5423,7 +5607,7 @@ ${mcpBudgetNote}` : "");
               const newMsgs2 = provider.buildToolResultMessages(result.toolCalls, syntheticResults, reasoningContent2);
               extraMessages.push(...newMsgs2);
               if (roundUsage.inputTokens > 0 || roundUsage.outputTokens > 0) {
-                this.addSessionUsage(roundUsage);
+                this.addSessionUsage(roundUsage, effectiveModel);
                 session.addTokenUsage(roundUsage);
                 if (teeShowTokens && !teeTokShown) {
                   this.renderer.renderUsage(roundUsage, this.sessionTokenUsage);
@@ -5438,14 +5622,14 @@ ${mcpBudgetNote}` : "");
         askUserContext.rl = this.rl;
         googleSearchContext.configManager = this.config;
         streamToFileContext.provider = provider;
-        streamToFileContext.model = this.currentModel;
+        streamToFileContext.model = effectiveModel;
         streamToFileContext.systemPrompt = systemPrompt;
         streamToFileContext.messages = apiMessages;
         streamToFileContext.extraMessages = extraMessages;
         streamToFileContext.temperature = modelParams.temperature;
         streamToFileContext.timeout = modelParams.timeout;
         spawnAgentContext.provider = provider;
-        spawnAgentContext.model = this.currentModel;
+        spawnAgentContext.model = effectiveModel;
         spawnAgentContext.systemPrompt = systemPrompt;
         spawnAgentContext.modelParams = modelParams;
         spawnAgentContext.configManager = this.config;
@@ -5605,7 +5789,7 @@ ${mcpBudgetNote}` : "");
         const summaryResult = await provider.chatWithTools(
           {
             messages: apiMessages,
-            model: this.currentModel,
+            model: effectiveModel,
             systemPrompt,
             stream: false,
             temperature: modelParams.temperature,
@@ -5643,7 +5827,7 @@ Tip: You can continue the conversation by asking the AI to proceed.`
         );
       }
       if (roundUsage.inputTokens > 0 || roundUsage.outputTokens > 0) {
-        this.addSessionUsage(roundUsage);
+        this.addSessionUsage(roundUsage, effectiveModel);
         session.addTokenUsage(roundUsage);
         if (this.shouldShowTokens()) {
           this.renderer.renderUsage(roundUsage, this.sessionTokenUsage);
@@ -5819,6 +6003,7 @@ Tip: You can continue the conversation by asking the AI to proceed.`
       forkSession: (messageCount, title) => this.sessions.forkSession(messageCount, title),
       getToolExecutor: () => this.toolExecutor,
       getCostTracker: () => this.costTracker,
+      computeRoutingDecision: (userInput) => this.computeRoutingDecision(userInput),
       exit: () => this.handleExit()
     };
     await cmd.execute(args, ctx);
@@ -5923,7 +6108,7 @@ program.command("web").description("Start Web UI server with browser-based chat
     console.error("Error: Invalid port number. Must be between 1 and 65535.");
     process.exit(1);
   }
-  const { startWebServer } = await import("./server-5HLNHRKM.js");
+  const { startWebServer } = await import("./server-MDBQX5UZ.js");
   await startWebServer({ port, host: options.host });
 });
 program.command("user [action] [username]").description("Manage Web UI users (list | create <name> | delete <name> | reset-password <name> | migrate <name>)").action(async (action, username) => {
@@ -6156,7 +6341,7 @@ program.command("hub [topic]").description("Start multi-agent hub (discuss / bra
     }),
     config.get("customProviders")
   );
-  const { startHub } = await import("./hub-RGK325NQ.js");
+  const { startHub } = await import("./hub-4VPTOMBP.js");
   await startHub(
     {
       topic: topic ?? "",

package/dist/{run-tests-KGUJL6L5.js → run-tests-OZ3OEOOB.js}

@@ -1,7 +1,7 @@
 import {
   executeTests,
   runTestsTool
-} from "./chunk-X7NVAEFI.js";
+} from "./chunk-VO5IZN2C.js";
 export {
   executeTests,
   runTestsTool

package/dist/{run-tests-NUF7CNM4.js → run-tests-WD53PYVA.js}

@@ -2,7 +2,7 @@
 import {
   executeTests,
   runTestsTool
-} from "./chunk-F44OKMB2.js";
+} from "./chunk-3LCVJ4AF.js";
 export {
   executeTests,
   runTestsTool

package/dist/{server-5HLNHRKM.js → server-MDBQX5UZ.js}

@@ -21,7 +21,7 @@ import {
   persistToolRound,
   rebuildExtraMessages,
   setupProxy
-} from "./chunk-XSIVGDCN.js";
+} from "./chunk-G5AISHJE.js";
 import {
   AuthManager
 } from "./chunk-BYNY5JPB.js";
@@ -42,7 +42,7 @@ import {
   spawnAgentContext,
   truncateOutput,
   undoStack
-} from "./chunk-BVLQ3FRA.js";
+} from "./chunk-Q5QSCO5D.js";
 import "./chunk-4BKXL7SM.js";
 import {
   AGENTIC_BEHAVIOR_GUIDELINE,
@@ -62,7 +62,7 @@ import {
   SKILLS_DIR_NAME,
   VERSION,
   buildUserIdentityPrompt
-} from "./chunk-F44OKMB2.js";
+} from "./chunk-3LCVJ4AF.js";
 
 // src/web/server.ts
 import express from "express";
@@ -1946,7 +1946,7 @@ ${undoResults.map((r) => `  \u2022 ${r}`).join("\n")}` });
       case "test": {
         this.send({ type: "info", message: "\u{1F9EA} Running tests..." });
         try {
-          const { executeTests } = await import("./run-tests-NUF7CNM4.js");
+          const { executeTests } = await import("./run-tests-WD53PYVA.js");
           const argStr = args.join(" ").trim();
           let testArgs = {};
           if (argStr) {

package/dist/{task-orchestrator-4PVBMQJ7.js → task-orchestrator-WDRXASIC.js}

@@ -4,11 +4,11 @@ import {
   getDangerLevel,
   googleSearchContext,
   truncateOutput
-} from "./chunk-BVLQ3FRA.js";
+} from "./chunk-Q5QSCO5D.js";
 import "./chunk-4BKXL7SM.js";
 import {
   SUBAGENT_ALLOWED_TOOLS
-} from "./chunk-F44OKMB2.js";
+} from "./chunk-3LCVJ4AF.js";
 
 // src/hub/task-orchestrator.ts
 import { createInterface } from "readline";

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "jinzd-ai-cli",
-  "version": "0.4.67",
+  "version": "0.4.68",
   "description": "Cross-platform REPL-style AI CLI with multi-provider support",
   "type": "module",
   "main": "./dist/index.js",