jinzd-ai-cli 0.4.53 → 0.4.55

package/dist/index.js

@@ -11,20 +11,23 @@ import {
   buildPhantomCorrectionMessage,
   buildWriteRoundReminder,
   clearDevState,
+  computeCost,
   detectsHallucinatedFileOp,
   extractWrittenFilePaths,
   findPhantomClaims,
+  formatCost,
   formatGitContextForPrompt,
   getContentText,
   getGitContext,
   getGitRoot,
+  getPricing,
   hadPreviousWriteToolCalls,
   loadDevState,
   parseSimpleYaml,
   saveDevState,
   sessionHasMeaningfulContent,
   setupProxy
-} from "./chunk-6I5FUNPR.js";
+} from "./chunk-JL5NK6AR.js";
 import {
   ToolExecutor,
   ToolRegistry,
@@ -38,7 +41,7 @@ import {
   spawnAgentContext,
   theme,
   undoStack
-} from "./chunk-IXDGWT2Z.js";
+} from "./chunk-YQEIQJ6K.js";
 import {
   fileCheckpoints
 } from "./chunk-4BKXL7SM.js";
@@ -63,7 +66,7 @@ import {
   SKILLS_DIR_NAME,
   VERSION,
   buildUserIdentityPrompt
-} from "./chunk-YIMTDKUW.js";
+} from "./chunk-W7QVBFIJ.js";
 
 // src/index.ts
 import { program } from "commander";
@@ -487,8 +490,12 @@ Error${typeName}: ${lines.join("\n")}
   renderUsage(usage, sessionTotal) {
     const total = usage.inputTokens + usage.outputTokens;
     let line = theme.dim("\u{1F4CA} ") + theme.dim(`in ${usage.inputTokens.toLocaleString()}`) + theme.dim(" + ") + theme.dim(`out ${usage.outputTokens.toLocaleString()}`) + theme.dim(` = ${total.toLocaleString()} tokens`);
+    const cacheRead = usage.cacheReadTokens ?? 0;
+    if (cacheRead > 0) {
+      line += theme.dim(`  \u2502  cache: ${cacheRead.toLocaleString()}`);
+    }
     if (sessionTotal) {
-      const sessionSum = sessionTotal.inputTokens + sessionTotal.outputTokens;
+      const sessionSum = sessionTotal.inputTokens + sessionTotal.outputTokens + (sessionTotal.cacheCreationTokens ?? 0) + (sessionTotal.cacheReadTokens ?? 0);
       line += theme.dim(`  \u2502  session total: ${sessionSum.toLocaleString()}`);
     }
     process.stdout.write(line + "\n\n");
@@ -1217,11 +1224,18 @@ function createDefaultCommands() {
         if (sys) {
           console.log(`  System   : ${sys.slice(0, 60)}...`);
         }
-        const totalTokens = tokenUsage.inputTokens + tokenUsage.outputTokens;
+        const cacheRead = tokenUsage.cacheReadTokens ?? 0;
+        const cacheCreate = tokenUsage.cacheCreationTokens ?? 0;
+        const totalTokens = tokenUsage.inputTokens + tokenUsage.outputTokens + cacheRead + cacheCreate;
         if (totalTokens > 0) {
+          const cacheSuffix = cacheRead > 0 || cacheCreate > 0 ? ` [cache: +${cacheCreate.toLocaleString()} / -${cacheRead.toLocaleString()}]` : "";
           console.log(
-            `  Tokens   : in ${tokenUsage.inputTokens.toLocaleString()} + out ${tokenUsage.outputTokens.toLocaleString()} = ${totalTokens.toLocaleString()} (session total)`
+            `  Tokens   : in ${tokenUsage.inputTokens.toLocaleString()} + out ${tokenUsage.outputTokens.toLocaleString()} = ${totalTokens.toLocaleString()}${cacheSuffix}`
           );
+          const cost = computeCost(ctx.getCurrentProvider(), ctx.getCurrentModel(), tokenUsage);
+          if (cost != null) {
+            console.log(`  Cost     : ${formatCost(cost)} (session total)`);
+          }
         }
         const ctxWindowSize = ctx.getContextWindowSize();
         if (ctxWindowSize > 0) {
@@ -1823,7 +1837,7 @@ ${hint}` : "")
     },
     {
       name: "cost",
-      description: "Show session token usage summary",
+      description: "Show session token usage, prompt-cache hits, and USD cost",
       usage: "/cost [reset]",
       execute(args, ctx) {
         const sub = args[0]?.toLowerCase();
@@ -1832,24 +1846,65 @@ ${hint}` : "")
           ctx.renderer.printSuccess("Session token counters reset.");
           return;
         }
-        const usage = ctx.getSessionTokenUsage();
-        const totalTokens = usage.inputTokens + usage.outputTokens;
+        const session = ctx.sessions.current;
+        const usage = session?.tokenUsage ?? {
+          inputTokens: 0,
+          outputTokens: 0,
+          cacheCreationTokens: 0,
+          cacheReadTokens: 0
+        };
+        const cacheCreate = usage.cacheCreationTokens ?? 0;
+        const cacheRead = usage.cacheReadTokens ?? 0;
+        const totalTokens = usage.inputTokens + usage.outputTokens + cacheCreate + cacheRead;
         if (totalTokens === 0) {
           ctx.renderer.printInfo("No token usage recorded this session.");
           return;
         }
+        const provider = ctx.getCurrentProvider();
+        const model = ctx.getCurrentModel();
+        const cost = computeCost(provider, model, usage);
+        const pricing = getPricing(provider, model);
+        let savings = null;
+        if (cost != null && pricing && cacheRead > 0) {
+          const costWithoutCache = computeCost(provider, model, {
+            inputTokens: usage.inputTokens + cacheRead,
+            outputTokens: usage.outputTokens,
+            cacheCreationTokens: cacheCreate,
+            cacheReadTokens: 0
+          });
+          if (costWithoutCache != null) savings = costWithoutCache - cost;
+        }
         console.log();
-        console.log(theme.heading("  \u{1F4CA} Session Token Usage"));
-        console.log(theme.dim("  " + "\u2500".repeat(40)));
-        console.log(theme.dim("  Input tokens  : ") + chalk2.white(usage.inputTokens.toLocaleString()));
-        console.log(theme.dim("  Output tokens : ") + chalk2.white(usage.outputTokens.toLocaleString()));
-        console.log(theme.dim("  Total tokens  : ") + chalk2.bold.white(totalTokens.toLocaleString()));
-        console.log(theme.dim("  " + "\u2500".repeat(40)));
-        const session = ctx.sessions.current;
+        console.log(theme.heading("  \u{1F4B0} Session Cost & Token Usage"));
+        console.log(theme.dim("  " + "\u2500".repeat(48)));
+        console.log(theme.dim("  Input (uncached) : ") + chalk2.white(usage.inputTokens.toLocaleString().padStart(12)));
+        console.log(theme.dim("  Output           : ") + chalk2.white(usage.outputTokens.toLocaleString().padStart(12)));
+        if (cacheCreate > 0) {
+          console.log(theme.dim("  Cache write      : ") + chalk2.yellow(cacheCreate.toLocaleString().padStart(12)));
+        }
+        if (cacheRead > 0) {
+          const pct = Math.round(cacheRead / (cacheRead + usage.inputTokens) * 100);
+          console.log(
+            theme.dim("  Cache read       : ") + chalk2.green(cacheRead.toLocaleString().padStart(12)) + theme.dim(`  (${pct}% hit rate)`)
+          );
+        }
+        console.log(theme.dim("  Total tokens     : ") + chalk2.bold.white(totalTokens.toLocaleString().padStart(12)));
+        console.log(theme.dim("  " + "\u2500".repeat(48)));
+        if (cost != null) {
+          console.log(theme.dim("  Cost             : ") + chalk2.bold.cyan(formatCost(cost).padStart(12)));
+          if (savings != null && savings > 0) {
+            console.log(
+              theme.dim("  Cache savings    : ") + chalk2.green(`-${formatCost(savings)}`.padStart(12)) + theme.dim(`  (vs no cache)`)
+            );
+          }
+        } else {
+          console.log(theme.dim("  Cost             : ") + theme.dim("      \u2014 (pricing unknown)"));
+        }
+        console.log(theme.dim("  " + "\u2500".repeat(48)));
         if (session) {
-          console.log(theme.dim("  Provider      : ") + theme.dim(ctx.getCurrentProvider()));
-          console.log(theme.dim("  Model         : ") + theme.dim(ctx.getCurrentModel()));
-          console.log(theme.dim("  Messages      : ") + theme.dim(String(session.messages.length)));
+          console.log(theme.dim("  Provider         : ") + theme.dim(provider));
+          console.log(theme.dim("  Model            : ") + theme.dim(model));
+          console.log(theme.dim("  Messages         : ") + theme.dim(String(session.messages.length)));
         }
         console.log();
       }
@@ -2106,7 +2161,7 @@ ${hint}` : "")
       usage: "/test [command|filter]",
       async execute(args, ctx) {
         try {
-          const { executeTests } = await import("./run-tests-3NNL7Z2E.js");
+          const { executeTests } = await import("./run-tests-X4PCLXA2.js");
           const argStr = args.join(" ").trim();
           let testArgs = {};
           if (argStr) {
@@ -3344,7 +3399,14 @@ var Repl = class {
   /** 当前加载的层级上下文（全局/项目/子目录） */
   contextLayers = [];
   /** 本次会话累计 token 用量 */
-  sessionTokenUsage = { inputTokens: 0, outputTokens: 0 };
+  sessionTokenUsage = { inputTokens: 0, outputTokens: 0, cacheCreationTokens: 0, cacheReadTokens: 0 };
+  /** Fold a single-request TokenUsage (with optional cache fields) into sessionTokenUsage. */
+  addSessionUsage(u) {
+    this.sessionTokenUsage.inputTokens += u.inputTokens;
+    this.sessionTokenUsage.outputTokens += u.outputTokens;
+    this.sessionTokenUsage.cacheCreationTokens += u.cacheCreationTokens ?? 0;
+    this.sessionTokenUsage.cacheReadTokens += u.cacheReadTokens ?? 0;
+  }
   /** 启动时检测到的 Git 分支（无 git 仓库时为 null） */
   gitBranch = null;
   /** MCP 多服务器管理器（无 MCP 配置时为 null） */
@@ -4607,8 +4669,7 @@ Session '${this.resumeSessionId}' not found.
         session.addMessage({ role: "assistant", content, timestamp: /* @__PURE__ */ new Date() });
         this.events.emit("message.after", { content });
         if (usage) {
-          this.sessionTokenUsage.inputTokens += usage.inputTokens;
-          this.sessionTokenUsage.outputTokens += usage.outputTokens;
+          this.addSessionUsage(usage);
           session.addTokenUsage(usage);
           if (showTokens && !tokensShown) {
             this.renderer.renderUsage(usage, this.sessionTokenUsage);
@@ -4637,8 +4698,7 @@ 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.sessionTokenUsage.inputTokens += response.usage.inputTokens;
-          this.sessionTokenUsage.outputTokens += response.usage.outputTokens;
+          this.addSessionUsage(response.usage);
           session.addTokenUsage(response.usage);
           if (this.shouldShowTokens()) {
             this.renderer.renderUsage(response.usage, this.sessionTokenUsage);
@@ -4797,7 +4857,7 @@ You have a maximum of ${maxToolRounds} tool call rounds for this task. Plan effi
     const modelParams = this.getModelParams();
     const useStreaming = this.config.get("ui").streaming;
     const spinner = this.renderer.showSpinner("Thinking...");
-    const roundUsage = { inputTokens: 0, outputTokens: 0 };
+    const roundUsage = { inputTokens: 0, outputTokens: 0, cacheCreationTokens: 0, cacheReadTokens: 0 };
     const supportsStreamingTools = useStreaming && typeof provider.chatWithToolsStream === "function";
     let consecutiveFreeRounds = 0;
     let lastToolCallSignature = "";
@@ -4904,6 +4964,8 @@ You have a maximum of ${maxToolRounds} tool call rounds for this task. Plan effi
         if (result.usage) {
           roundUsage.inputTokens += result.usage.inputTokens;
           roundUsage.outputTokens += result.usage.outputTokens;
+          roundUsage.cacheCreationTokens += result.usage.cacheCreationTokens ?? 0;
+          roundUsage.cacheReadTokens += result.usage.cacheReadTokens ?? 0;
         }
         if ("content" in result) {
           const hasWriteTools = toolDefs.some((t) => t.name === "write_file" || t.name === "edit_file");
@@ -4954,8 +5016,7 @@ You have a maximum of ${maxToolRounds} tool call rounds for this task. Plan effi
           });
           this.events.emit("message.after", { content: finalContent });
           if (roundUsage.inputTokens > 0 || roundUsage.outputTokens > 0) {
-            this.sessionTokenUsage.inputTokens += roundUsage.inputTokens;
-            this.sessionTokenUsage.outputTokens += roundUsage.outputTokens;
+            this.addSessionUsage(roundUsage);
             session.addTokenUsage(roundUsage);
             if (this.shouldShowTokens()) {
               this.renderer.renderUsage(roundUsage, this.sessionTokenUsage);
@@ -4993,6 +5054,8 @@ You have a maximum of ${maxToolRounds} tool call rounds for this task. Plan effi
               if (genUsage) {
                 roundUsage.inputTokens += genUsage.inputTokens;
                 roundUsage.outputTokens += genUsage.outputTokens;
+                roundUsage.cacheCreationTokens += genUsage.cacheCreationTokens ?? 0;
+                roundUsage.cacheReadTokens += genUsage.cacheReadTokens ?? 0;
               }
               session.addMessage({ role: "assistant", content: genContent, timestamp: /* @__PURE__ */ new Date() });
               this.events.emit("message.after", { content: genContent });
@@ -5007,8 +5070,7 @@ You have a maximum of ${maxToolRounds} tool call rounds for this task. Plan effi
               const newMsgs2 = provider.buildToolResultMessages(result.toolCalls, syntheticResults, reasoningContent2);
               extraMessages.push(...newMsgs2);
               if (roundUsage.inputTokens > 0 || roundUsage.outputTokens > 0) {
-                this.sessionTokenUsage.inputTokens += roundUsage.inputTokens;
-                this.sessionTokenUsage.outputTokens += roundUsage.outputTokens;
+                this.addSessionUsage(roundUsage);
                 session.addTokenUsage(roundUsage);
                 if (teeShowTokens && !teeTokShown) {
                   this.renderer.renderUsage(roundUsage, this.sessionTokenUsage);
@@ -5204,6 +5266,8 @@ You have a maximum of ${maxToolRounds} tool call rounds for this task. Plan effi
           if (summaryResult.usage) {
             roundUsage.inputTokens += summaryResult.usage.inputTokens;
             roundUsage.outputTokens += summaryResult.usage.outputTokens;
+            roundUsage.cacheCreationTokens += summaryResult.usage.cacheCreationTokens ?? 0;
+            roundUsage.cacheReadTokens += summaryResult.usage.cacheReadTokens ?? 0;
           }
         } else {
           this.renderer.renderError(
@@ -5218,8 +5282,7 @@ Tip: You can continue the conversation by asking the AI to proceed.`
         );
       }
       if (roundUsage.inputTokens > 0 || roundUsage.outputTokens > 0) {
-        this.sessionTokenUsage.inputTokens += roundUsage.inputTokens;
-        this.sessionTokenUsage.outputTokens += roundUsage.outputTokens;
+        this.addSessionUsage(roundUsage);
         session.addTokenUsage(roundUsage);
         if (this.shouldShowTokens()) {
           this.renderer.renderUsage(roundUsage, this.sessionTokenUsage);
@@ -5316,7 +5379,12 @@ Tip: You can continue the conversation by asking the AI to proceed.`
       },
       getSessionTokenUsage: () => ({ ...this.sessionTokenUsage }),
       resetSessionTokenUsage: () => {
-        this.sessionTokenUsage = { inputTokens: 0, outputTokens: 0 };
+        this.sessionTokenUsage = {
+          inputTokens: 0,
+          outputTokens: 0,
+          cacheCreationTokens: 0,
+          cacheReadTokens: 0
+        };
       },
       getGitBranch: () => this.gitBranch,
       getLastResponse: () => lastResponseStore.content,
@@ -5493,7 +5561,7 @@ 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-PFHWO3HL.js");
+  const { startWebServer } = await import("./server-YPAZWGUE.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) => {
@@ -5726,7 +5794,7 @@ program.command("hub [topic]").description("Start multi-agent hub (discuss / bra
     }),
     config.get("customProviders")
   );
-  const { startHub } = await import("./hub-4DNFD6JK.js");
+  const { startHub } = await import("./hub-AUWP4SWJ.js");
   await startHub(
     {
       topic: topic ?? "",

package/dist/{run-tests-NJQK4B43.js → run-tests-I6UDHVIS.js}

@@ -1,7 +1,7 @@
 import {
   executeTests,
   runTestsTool
-} from "./chunk-W6AK76UM.js";
+} from "./chunk-DJ342VFS.js";
 export {
   executeTests,
   runTestsTool

package/dist/{run-tests-3NNL7Z2E.js → run-tests-X4PCLXA2.js}

@@ -2,7 +2,7 @@
 import {
   executeTests,
   runTestsTool
-} from "./chunk-YIMTDKUW.js";
+} from "./chunk-W7QVBFIJ.js";
 export {
   executeTests,
   runTestsTool

package/dist/{server-PFHWO3HL.js → server-YPAZWGUE.js}

@@ -7,7 +7,9 @@ import {
   SessionManager,
   SkillManager,
   TOOL_CALL_REMINDER,
+  computeCost,
   detectsHallucinatedFileOp,
+  formatCost,
   formatGitContextForPrompt,
   getContentText,
   getGitContext,
@@ -15,7 +17,7 @@ import {
   hadPreviousWriteToolCalls,
   loadDevState,
   setupProxy
-} from "./chunk-6I5FUNPR.js";
+} from "./chunk-JL5NK6AR.js";
 import {
   AuthManager
 } from "./chunk-BYNY5JPB.js";
@@ -34,7 +36,7 @@ import {
   spawnAgentContext,
   truncateOutput,
   undoStack
-} from "./chunk-IXDGWT2Z.js";
+} from "./chunk-YQEIQJ6K.js";
 import "./chunk-4BKXL7SM.js";
 import {
   AGENTIC_BEHAVIOR_GUIDELINE,
@@ -54,7 +56,7 @@ import {
   SKILLS_DIR_NAME,
   VERSION,
   buildUserIdentityPrompt
-} from "./chunk-YIMTDKUW.js";
+} from "./chunk-W7QVBFIJ.js";
 
 // src/web/server.ts
 import express from "express";
@@ -483,7 +485,17 @@ var SessionHandler = class _SessionHandler {
   currentModel;
   planMode = false;
   runtimeThinking = null;
-  sessionTokenUsage = { inputTokens: 0, outputTokens: 0 };
+  sessionTokenUsage = { inputTokens: 0, outputTokens: 0, cacheCreationTokens: 0, cacheReadTokens: 0 };
+  /** Accumulate a TokenUsage (with optional cache fields) into sessionTokenUsage. */
+  addWebSessionUsage(u) {
+    this.sessionTokenUsage.inputTokens += u.inputTokens;
+    this.sessionTokenUsage.outputTokens += u.outputTokens;
+    this.sessionTokenUsage.cacheCreationTokens += u.cacheCreationTokens ?? 0;
+    this.sessionTokenUsage.cacheReadTokens += u.cacheReadTokens ?? 0;
+  }
+  resetWebSessionUsage() {
+    this.sessionTokenUsage = { inputTokens: 0, outputTokens: 0, cacheCreationTokens: 0, cacheReadTokens: 0 };
+  }
   abortController = null;
   userInterjection = null;
   processing = false;
@@ -547,6 +559,7 @@ var SessionHandler = class _SessionHandler {
       displayName: p.info.displayName,
       models: p.info.models.map((m) => ({ id: m.id, name: m.displayName ?? m.id }))
     }));
+    const costUsd = computeCost(this.currentProvider, this.currentModel, this.sessionTokenUsage);
     this.send({
       type: "status",
       provider: this.currentProvider,
@@ -557,6 +570,7 @@ var SessionHandler = class _SessionHandler {
       planMode: this.planMode,
       thinkingMode: this.runtimeThinking ?? false,
       tokenUsage: { ...this.sessionTokenUsage },
+      costUsd,
       providers: providerList
     });
   }
@@ -724,8 +738,7 @@ var SessionHandler = class _SessionHandler {
         if (chunk.done) {
           this.send({ type: "response_done", content: fullContent, usage: chunk.usage });
           if (chunk.usage) {
-            this.sessionTokenUsage.inputTokens += chunk.usage.inputTokens;
-            this.sessionTokenUsage.outputTokens += chunk.usage.outputTokens;
+            this.addWebSessionUsage(chunk.usage);
             session.addTokenUsage(chunk.usage);
           }
           break;
@@ -759,7 +772,7 @@ You have a maximum of ${maxToolRounds} tool call rounds for this task. Plan effi
 - When remaining rounds are low, focus on completing the current task and summarizing.`;
     const systemPrompt = baseSystemPrompt + roundBudgetHint;
     const modelParams = this.getModelParams();
-    const roundUsage = { inputTokens: 0, outputTokens: 0 };
+    const roundUsage = { inputTokens: 0, outputTokens: 0, cacheCreationTokens: 0, cacheReadTokens: 0 };
     const supportsStreamingTools = typeof provider.chatWithToolsStream === "function";
     let consecutiveFreeRounds = 0;
     const warnNoteAt = Math.max(10, Math.floor(maxToolRounds * 0.2));
@@ -829,6 +842,8 @@ You have a maximum of ${maxToolRounds} tool call rounds for this task. Plan effi
         if (result.usage) {
           roundUsage.inputTokens += result.usage.inputTokens;
           roundUsage.outputTokens += result.usage.outputTokens;
+          roundUsage.cacheCreationTokens += result.usage.cacheCreationTokens ?? 0;
+          roundUsage.cacheReadTokens += result.usage.cacheReadTokens ?? 0;
         }
         if (result.content && !result.toolCalls) {
           const hasWriteTools = toolDefs.some((t) => t.name === "write_file" || t.name === "edit_file");
@@ -843,8 +858,7 @@ You have a maximum of ${maxToolRounds} tool call rounds for this task. Plan effi
           }
           this.send({ type: "response_done", content: result.content, usage: roundUsage });
           session.addMessage({ role: "assistant", content: result.content, timestamp: /* @__PURE__ */ new Date() });
-          this.sessionTokenUsage.inputTokens += roundUsage.inputTokens;
-          this.sessionTokenUsage.outputTokens += roundUsage.outputTokens;
+          this.addWebSessionUsage(roundUsage);
           session.addTokenUsage(roundUsage);
           return;
         }
@@ -917,8 +931,7 @@ ${summaryResult.content}`,
           message: `Reached maximum tool call rounds (${maxToolRounds}). You can continue by asking the AI to proceed.`
         });
       }
-      this.sessionTokenUsage.inputTokens += roundUsage.inputTokens;
-      this.sessionTokenUsage.outputTokens += roundUsage.outputTokens;
+      this.addWebSessionUsage(roundUsage);
       session.addTokenUsage(roundUsage);
     } catch (err) {
       if (err.name === "AbortError") {
@@ -1034,7 +1047,7 @@ ${summaryResult.content}`,
       case "clear":
         this.saveIfNeeded();
         this.sessions.createSession(this.currentProvider, this.currentModel);
-        this.sessionTokenUsage = { inputTokens: 0, outputTokens: 0 };
+        this.resetWebSessionUsage();
         this.send({ type: "info", message: "Conversation cleared." });
         this.sendStatus();
         this.sendSessionList();
@@ -1068,12 +1081,19 @@ ${summaryResult.content}`,
       }
       case "status": {
         const session = this.sessions.current;
+        const cacheRead = this.sessionTokenUsage.cacheReadTokens;
+        const cacheCreate = this.sessionTokenUsage.cacheCreationTokens;
+        const cost = computeCost(this.currentProvider, this.currentModel, this.sessionTokenUsage);
+        const cacheLine = cacheRead > 0 || cacheCreate > 0 ? `
+Cache:  write=${cacheCreate} read=${cacheRead}` : "";
+        const costLine = cost != null ? `
+Cost:   ${formatCost(cost)}` : "";
         this.send({
           type: "info",
           message: `Provider: ${this.currentProvider}
 Model: ${this.currentModel}
 Session: ${session?.id ?? "none"} (${session?.messages.length ?? 0} messages)
-Tokens: in=${this.sessionTokenUsage.inputTokens} out=${this.sessionTokenUsage.outputTokens}`
+Tokens: in=${this.sessionTokenUsage.inputTokens} out=${this.sessionTokenUsage.outputTokens}${cacheLine}${costLine}`
         });
         break;
       }
@@ -1083,7 +1103,7 @@ Tokens: in=${this.sessionTokenUsage.inputTokens} out=${this.sessionTokenUsage.ou
           this.saveIfNeeded();
           const created = this.sessions.createSession(this.currentProvider, this.currentModel);
           this.unsavedSessions.set(created.id, created);
-          this.sessionTokenUsage = { inputTokens: 0, outputTokens: 0 };
+          this.resetWebSessionUsage();
           this.send({ type: "info", message: "New session created." });
           this.sendStatus();
           this.sendSessionList();
@@ -1094,7 +1114,7 @@ Tokens: in=${this.sessionTokenUsage.inputTokens} out=${this.sessionTokenUsage.ou
           const cached = cachedExact ?? [...this.unsavedSessions.values()].find((s) => s.id.startsWith(targetId));
           if (cached) {
             this.sessions.setCurrent(cached);
-            this.sessionTokenUsage = { inputTokens: 0, outputTokens: 0 };
+            this.resetWebSessionUsage();
             this.send({
               type: "info",
               message: `Loaded session: ${cached.id.slice(0, 8)} "${cached.title ?? ""}" (${cached.messages.length} messages)`
@@ -1108,7 +1128,7 @@ Tokens: in=${this.sessionTokenUsage.inputTokens} out=${this.sessionTokenUsage.ou
           const found = list.find((s) => s.id.startsWith(targetId));
           if (found) {
             this.sessions.loadSession(found.id);
-            this.sessionTokenUsage = { inputTokens: 0, outputTokens: 0 };
+            this.resetWebSessionUsage();
             this.send({ type: "info", message: `Loaded session: ${found.id.slice(0, 8)} "${found.title ?? ""}" (${found.messageCount} messages)` });
             this.sendSessionMessages();
             this.sendStatus();
@@ -1116,7 +1136,7 @@ Tokens: in=${this.sessionTokenUsage.inputTokens} out=${this.sessionTokenUsage.ou
           } else {
             const recreated = this.sessions.createSession(this.currentProvider, this.currentModel);
             this.unsavedSessions.set(recreated.id, recreated);
-            this.sessionTokenUsage = { inputTokens: 0, outputTokens: 0 };
+            this.resetWebSessionUsage();
             this.send({
               type: "info",
               message: `Previous session (${targetId.slice(0, 8)}) is no longer available \u2014 started a new one.`
@@ -1256,16 +1276,41 @@ Tokens: in=${this.sessionTokenUsage.inputTokens} out=${this.sessionTokenUsage.ou
         });
         break;
       case "cost": {
-        const total = this.sessionTokenUsage.inputTokens + this.sessionTokenUsage.outputTokens;
-        this.send({
-          type: "info",
-          message: `\u{1F4CA} Token Usage
-  Provider: ${this.currentProvider}
-  Model: ${this.currentModel}
-  Input:  ${this.sessionTokenUsage.inputTokens.toLocaleString()}
-  Output: ${this.sessionTokenUsage.outputTokens.toLocaleString()}
-  Total:  ${total.toLocaleString()}`
-        });
+        const u = this.sessionTokenUsage;
+        const total = u.inputTokens + u.outputTokens + u.cacheCreationTokens + u.cacheReadTokens;
+        const cost = computeCost(this.currentProvider, this.currentModel, u);
+        let savings = null;
+        if (cost != null && u.cacheReadTokens > 0) {
+          const withoutCache = computeCost(this.currentProvider, this.currentModel, {
+            inputTokens: u.inputTokens + u.cacheReadTokens,
+            outputTokens: u.outputTokens,
+            cacheCreationTokens: u.cacheCreationTokens,
+            cacheReadTokens: 0
+          });
+          if (withoutCache != null) savings = withoutCache - cost;
+        }
+        const lines = [
+          "\u{1F4B0} Session Cost & Token Usage",
+          `  Provider        : ${this.currentProvider}`,
+          `  Model           : ${this.currentModel}`,
+          `  Input (uncached): ${u.inputTokens.toLocaleString()}`,
+          `  Output          : ${u.outputTokens.toLocaleString()}`
+        ];
+        if (u.cacheCreationTokens > 0) lines.push(`  Cache write     : ${u.cacheCreationTokens.toLocaleString()}`);
+        if (u.cacheReadTokens > 0) {
+          const pct = Math.round(u.cacheReadTokens / (u.cacheReadTokens + u.inputTokens) * 100);
+          lines.push(`  Cache read      : ${u.cacheReadTokens.toLocaleString()} (${pct}% hit rate)`);
+        }
+        lines.push(`  Total tokens    : ${total.toLocaleString()}`);
+        if (cost != null) {
+          lines.push(`  Cost            : ${formatCost(cost)}`);
+          if (savings != null && savings > 0) {
+            lines.push(`  Cache savings   : -${formatCost(savings)} (vs no cache)`);
+          }
+        } else {
+          lines.push(`  Cost            : \u2014 (pricing unknown for this model)`);
+        }
+        this.send({ type: "info", message: lines.join("\n") });
         break;
       }
       case "tools":
@@ -1691,7 +1736,7 @@ ${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-3NNL7Z2E.js");
+          const { executeTests } = await import("./run-tests-X4PCLXA2.js");
           const argStr = args.join(" ").trim();
           let testArgs = {};
           if (argStr) {
@@ -2215,6 +2260,7 @@ Add .md files to create commands.` });
     }));
     this.send({
       type: "session_messages",
+      sessionId: session.id,
       messages
     });
   }

package/dist/{task-orchestrator-C42TNHE6.js → task-orchestrator-MWO6A4KQ.js}

@@ -4,11 +4,11 @@ import {
   getDangerLevel,
   googleSearchContext,
   truncateOutput
-} from "./chunk-IXDGWT2Z.js";
+} from "./chunk-YQEIQJ6K.js";
 import "./chunk-4BKXL7SM.js";
 import {
   SUBAGENT_ALLOWED_TOOLS
-} from "./chunk-YIMTDKUW.js";
+} from "./chunk-W7QVBFIJ.js";
 
 // src/hub/task-orchestrator.ts
 import { createInterface } from "readline";