@bike4mind/cli 0.2.30-feat-universal-prompt-caching.19181 → 0.2.30-feat-cli-scrollable-command-autocomplete.19203

package/dist/{artifactExtractor-FPTQA6QK.js → artifactExtractor-SVFEWL53.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import {
   CurationArtifactType
-} from "./chunk-LQRI6AKT.js";
+} from "./chunk-C3K7SH2V.js";
 
 // ../../b4m-core/packages/services/dist/src/notebookCurationService/artifactExtractor.js
 var ARTIFACT_TAG_REGEX = /<artifact\s+(.*?)>([\s\S]*?)<\/artifact>/gi;

package/dist/{chunk-7FGWZUYS.js → chunk-5HA4JEOV.js}

@@ -7,6 +7,7 @@ import {
   ModelBackend,
   OpenAIEmbeddingModel,
   PermissionDeniedError,
+  REASONING_SUPPORTED_MODELS,
   SpeechToTextModels,
   SupportedFabFileMimeTypes,
   VIDEO_SIZE_CONSTRAINTS,
@@ -15,7 +16,7 @@ import {
   dayjsConfig_default,
   extractSnippetMeta,
   settingsMap
-} from "./chunk-LQRI6AKT.js";
+} from "./chunk-C3K7SH2V.js";
 import {
   Logger
 } from "./chunk-OCYRD7D6.js";
@@ -2086,212 +2087,6 @@ function getLastBuildDebugInfo() {
   return buildAndSortMessages.lastDebugInfo?.messageTruncation || null;
 }
 
-// ../../b4m-core/packages/utils/dist/src/llm/caching/adapters/anthropic.js
-var AnthropicCachingAdapter = class {
-  applyCaching(apiParams, strategy) {
-    if (!strategy.enableCaching)
-      return apiParams;
-    const ttl = strategy.cacheTTL ?? "5m";
-    const modifiedParams = { ...apiParams };
-    if (strategy.cacheTools && modifiedParams.tools?.length > 0) {
-      const tools = [...modifiedParams.tools];
-      const lastTool = tools[tools.length - 1];
-      tools[tools.length - 1] = {
-        ...lastTool,
-        cache_control: { type: "ephemeral", ...ttl === "1h" ? { ttl } : {} }
-      };
-      modifiedParams.tools = tools;
-    }
-    if (strategy.cacheSystemPrompt && modifiedParams.system?.length > 0) {
-      const system = Array.isArray(modifiedParams.system) ? [...modifiedParams.system] : [{ type: "text", text: modifiedParams.system }];
-      const lastBlock = system[system.length - 1];
-      system[system.length - 1] = {
-        ...lastBlock,
-        cache_control: { type: "ephemeral", ...ttl === "1h" ? { ttl } : {} }
-      };
-      modifiedParams.system = system;
-    }
-    if (strategy.cacheConversationHistory && modifiedParams.messages?.length > 0) {
-      const messages = [...modifiedParams.messages];
-      const lastMsg = messages[messages.length - 1];
-      let content = lastMsg.content;
-      if (typeof content === "string") {
-        content = [{ type: "text", text: content }];
-      } else if (!Array.isArray(content)) {
-        return modifiedParams;
-      }
-      const contentArr = [...content];
-      const lastBlock = contentArr[contentArr.length - 1];
-      contentArr[contentArr.length - 1] = {
-        ...lastBlock,
-        cache_control: { type: "ephemeral", ...ttl === "1h" ? { ttl } : {} }
-      };
-      messages[messages.length - 1] = {
-        ...lastMsg,
-        content: contentArr
-      };
-      modifiedParams.messages = messages;
-    }
-    return modifiedParams;
-  }
-  extractCacheStats(response, model) {
-    const usage = response.usage;
-    if (!usage)
-      return void 0;
-    const cacheReadTokens = usage.cache_read_input_tokens || 0;
-    const cacheWriteTokens = usage.cache_creation_input_tokens || 0;
-    const uncachedTokens = usage.input_tokens || 0;
-    const totalInputTokens = cacheReadTokens + cacheWriteTokens + uncachedTokens;
-    const cacheHitRate = totalInputTokens > 0 ? cacheReadTokens / totalInputTokens * 100 : 0;
-    const costSavingsPercent = cacheHitRate * 0.9;
-    const estimatedLatencyReduction = cacheHitRate * 0.85;
-    return {
-      provider: ModelBackend.Anthropic,
-      model,
-      totalInputTokens,
-      cacheReadTokens,
-      cacheWriteTokens,
-      uncachedTokens,
-      cacheHitRate,
-      costSavingsPercent,
-      estimatedLatencyReduction,
-      providerMetadata: {
-        ttl: usage.cache_creation_input_tokens ? usage.cache_creation_input_tokens > 0 ? "5m" : "1h" : void 0
-      }
-    };
-  }
-};
-
-// ../../b4m-core/packages/utils/dist/src/llm/caching/adapters/openai.js
-var OpenAICachingAdapter = class {
-  applyCaching(apiParams, strategy) {
-    return apiParams;
-  }
-  extractCacheStats(response, model) {
-    const usage = response.usage;
-    if (!usage)
-      return void 0;
-    const totalInputTokens = usage.prompt_tokens || 0;
-    const cachedTokens = usage.prompt_tokens_details?.cached_tokens || 0;
-    const cacheHitRate = totalInputTokens > 0 ? cachedTokens / totalInputTokens * 100 : 0;
-    const costSavingsPercent = cacheHitRate * 0.9;
-    const estimatedLatencyReduction = cacheHitRate * 0.8;
-    return {
-      provider: ModelBackend.OpenAI,
-      model,
-      totalInputTokens,
-      cacheReadTokens: cachedTokens,
-      cacheWriteTokens: 0,
-      // OpenAI doesn't expose this
-      uncachedTokens: totalInputTokens - cachedTokens,
-      cacheHitRate,
-      costSavingsPercent,
-      estimatedLatencyReduction,
-      providerMetadata: {
-        automatic: true
-      }
-    };
-  }
-};
-
-// ../../b4m-core/packages/utils/dist/src/llm/caching/adapters/gemini.js
-var GeminiCachingAdapter = class {
-  applyCaching(apiParams, strategy) {
-    return apiParams;
-  }
-  extractCacheStats(response, model) {
-    const usage = response.usageMetadata;
-    if (!usage)
-      return void 0;
-    const totalInputTokens = usage.promptTokenCount || 0;
-    const cachedTokens = usage.cachedContentTokenCount || 0;
-    const cacheHitRate = totalInputTokens > 0 ? cachedTokens / totalInputTokens * 100 : 0;
-    const savingsMultiplier = model.includes("2.5") ? 0.9 : 0.75;
-    const costSavingsPercent = cacheHitRate * savingsMultiplier;
-    const estimatedLatencyReduction = cacheHitRate * 0.75;
-    return {
-      provider: ModelBackend.Gemini,
-      model,
-      totalInputTokens,
-      cacheReadTokens: cachedTokens,
-      cacheWriteTokens: 0,
-      // Not exposed
-      uncachedTokens: totalInputTokens - cachedTokens,
-      cacheHitRate,
-      costSavingsPercent,
-      estimatedLatencyReduction,
-      providerMetadata: {
-        implicit: true
-      }
-    };
-  }
-};
-
-// ../../b4m-core/packages/utils/dist/src/llm/caching/adapters/xai.js
-var XAICachingAdapter = class {
-  applyCaching(apiParams, strategy) {
-    return apiParams;
-  }
-  getHeaders(strategy) {
-    const headers = {};
-    if (strategy.conversationId) {
-      headers["x-grok-conv-id"] = strategy.conversationId;
-    }
-    return headers;
-  }
-  extractCacheStats(response, model) {
-    const usage = response.usage;
-    if (!usage)
-      return void 0;
-    const totalInputTokens = usage.prompt_tokens || 0;
-    const cachedTokens = usage.cached_prompt_tokens || 0;
-    const cacheHitRate = totalInputTokens > 0 ? cachedTokens / totalInputTokens * 100 : 0;
-    const costSavingsPercent = cacheHitRate * 0.65;
-    const estimatedLatencyReduction = cacheHitRate * 0.7;
-    return {
-      provider: ModelBackend.XAI,
-      model,
-      totalInputTokens,
-      cacheReadTokens: cachedTokens,
-      cacheWriteTokens: 0,
-      // Not exposed
-      uncachedTokens: totalInputTokens - cachedTokens,
-      cacheHitRate,
-      costSavingsPercent,
-      estimatedLatencyReduction,
-      providerMetadata: {
-        automatic: true,
-        conversationId: usage.x_grok_conv_id
-      }
-    };
-  }
-};
-
-// ../../b4m-core/packages/utils/dist/src/llm/caching/adapters/index.js
-var NoOpCachingAdapter = class {
-  applyCaching(apiParams) {
-    return apiParams;
-  }
-  extractCacheStats() {
-    return void 0;
-  }
-};
-var ADAPTERS = {
-  [ModelBackend.Anthropic]: new AnthropicCachingAdapter(),
-  [ModelBackend.OpenAI]: new OpenAICachingAdapter(),
-  [ModelBackend.Gemini]: new GeminiCachingAdapter(),
-  [ModelBackend.Bedrock]: new AnthropicCachingAdapter(),
-  // Uses Anthropic format
-  [ModelBackend.XAI]: new XAICachingAdapter(),
-  [ModelBackend.Ollama]: new NoOpCachingAdapter(),
-  [ModelBackend.BFL]: new NoOpCachingAdapter(),
-  [ModelBackend.VoyageAI]: new NoOpCachingAdapter(),
-  [ModelBackend.AWS]: new NoOpCachingAdapter()
-};
-function getCachingAdapter(backend) {
-  return ADAPTERS[backend] || new NoOpCachingAdapter();
-}
-
 // ../../b4m-core/packages/utils/dist/src/llm/anthropicBackend.js
 var TEMPERATURE_ONLY_MODELS = [
   ChatModels.CLAUDE_4_5_SONNET,
@@ -2791,19 +2586,6 @@ var AnthropicBackend = class {
       this.isThinkingEnabled = true;
       this.logger.debug(`[AnthropicBackend] QuestMaster enabled thinking with budget: ${thinkingBudget}, max_tokens: ${apiParams.max_tokens}, temperature: ${apiParams.temperature}, top_p: ${apiParams.top_p}`);
     }
-    const cacheStrategy = options.cacheStrategy;
-    if (cacheStrategy?.enableCaching) {
-      const adapter = getCachingAdapter(ModelBackend.Anthropic);
-      const cachedParams = adapter.applyCaching(apiParams, cacheStrategy);
-      Object.assign(apiParams, cachedParams);
-      this.logger.info("[PromptCache] Caching enabled for Anthropic request", {
-        model,
-        cacheSystemPrompt: cacheStrategy.cacheSystemPrompt,
-        cacheTools: cacheStrategy.cacheTools,
-        cacheConversationHistory: cacheStrategy.cacheConversationHistory,
-        cacheTTL: cacheStrategy.cacheTTL
-      });
-    }
     try {
       const func = [];
       if (options.stream) {
@@ -2849,7 +2631,6 @@ var AnthropicBackend = class {
                 clearTimeout(requestTimeout);
               let isInThinkingBlock = false;
               const collectedContent = [];
-              let usageInfo = void 0;
               const enableIdleTimeout = options._internal?.enableIdleTimeout ?? false;
               const configuredIdleTimeoutMs = options._internal?.idleTimeoutMs;
               const idleTimeoutMs = this.isThinkingEnabled ? THINKING_IDLE_TIMEOUT_MS : configuredIdleTimeoutMs ?? DEFAULT_IDLE_TIMEOUT_MS;
@@ -2951,10 +2732,6 @@ var AnthropicBackend = class {
                     func[0].name = toolEvent.name;
                     func[0].parameters = JSON.stringify(toolEvent.input);
                     func[0].id = toolEvent.id;
-                  } else if (event.type === "message_delta") {
-                    if ("usage" in event) {
-                      usageInfo = event.usage;
-                    }
                   }
                 }
               } finally {
@@ -2964,21 +2741,6 @@ var AnthropicBackend = class {
               if (this.isThinkingEnabled && collectedContent.length > 0) {
                 this.lastAssistantContent = collectedContent.filter((c) => c != null);
               }
-              let cacheStats;
-              if (cacheStrategy?.enableCaching && usageInfo) {
-                const adapter = getCachingAdapter(ModelBackend.Anthropic);
-                cacheStats = adapter.extractCacheStats({ usage: usageInfo }, model);
-                if (cacheStats) {
-                  this.logger.info("[PromptCache] Anthropic cache stats", {
-                    model,
-                    cacheHitRate: cacheStats.cacheHitRate.toFixed(2) + "%",
-                    costSavingsPercent: cacheStats.costSavingsPercent.toFixed(2) + "%",
-                    totalInputTokens: cacheStats.totalInputTokens,
-                    cacheReadTokens: cacheStats.cacheReadTokens,
-                    cacheWriteTokens: cacheStats.cacheWriteTokens
-                  });
-                }
-              }
               const hasToolsToExecute = func.some((f) => f && f.name);
               if (!hasToolsToExecute) {
                 if (options.tools && options.tools.length > 0) {
@@ -2988,7 +2750,7 @@ var AnthropicBackend = class {
                     this.logger.warn(`\u26A0\uFE0F [AnthropicBackend] Model returned empty content with ${mcpToolCount} MCP tools available. Total tools: ${options.tools.length}. This may indicate tool overload or model confusion.`);
                   }
                 }
-                await cb([], { toolsUsed, cacheStats });
+                await cb([], { toolsUsed });
               }
               resolve();
             } catch (error) {
@@ -4150,19 +3912,6 @@ var AnthropicBedrockBackend = class extends BaseBedrockBackend {
         topP: options.topP
       }
     }, null, 2)}`);
-    const cacheStrategy = options.cacheStrategy;
-    if (cacheStrategy?.enableCaching) {
-      const adapter = getCachingAdapter(ModelBackend.Bedrock);
-      const cachedBody = adapter.applyCaching(body, cacheStrategy);
-      Object.assign(body, cachedBody);
-      console.log("[PromptCache] Bedrock caching enabled", {
-        model: modelId,
-        cacheSystemPrompt: cacheStrategy.cacheSystemPrompt,
-        cacheTools: cacheStrategy.cacheTools,
-        cacheConversationHistory: cacheStrategy.cacheConversationHistory,
-        cacheTTL: cacheStrategy.cacheTTL
-      });
-    }
     const bodyStr = JSON.stringify(body);
     console.log(`[AnthropicBedrockBackend] Request body: ${bodyStr.substring(0, 1e3)}${bodyStr.length > 1e3 ? "..." : ""}`);
     return {
@@ -5663,26 +5412,10 @@ var GeminiBackend = class {
               });
             }
           });
-          const cacheStrategy = options.cacheStrategy;
-          let cacheStats;
-          if (cacheStrategy?.enableCaching && lastChunk.usageMetadata) {
-            const adapter = getCachingAdapter(ModelBackend.Gemini);
-            cacheStats = adapter.extractCacheStats(lastChunk, modelName);
-            if (cacheStats && cacheStats.cacheReadTokens > 0) {
-              this.logger.info("[PromptCache] Gemini cache stats (streaming)", {
-                model: modelName,
-                cacheHitRate: cacheStats.cacheHitRate.toFixed(2) + "%",
-                costSavingsPercent: cacheStats.costSavingsPercent.toFixed(2) + "%",
-                totalInputTokens: cacheStats.totalInputTokens,
-                cacheReadTokens: cacheStats.cacheReadTokens
-              });
-            }
-          }
           await callback([], {
             inputTokens: lastChunk.usageMetadata?.promptTokenCount ?? 0,
             outputTokens: lastChunk.usageMetadata?.candidatesTokenCount ?? 0,
-            toolsUsed,
-            cacheStats
+            toolsUsed
           });
         }
         if (toolCalls.length > 0 && options.tools?.length) {
@@ -5775,26 +5508,10 @@ var GeminiBackend = class {
         if (options.abortSignal?.aborted) {
           throw new Error("Request aborted");
         }
-        const cacheStrategy = options.cacheStrategy;
-        let cacheStats;
-        if (cacheStrategy?.enableCaching && r.usageMetadata) {
-          const adapter = getCachingAdapter(ModelBackend.Gemini);
-          cacheStats = adapter.extractCacheStats(r, modelName);
-          if (cacheStats && cacheStats.cacheReadTokens > 0) {
-            this.logger.info("[PromptCache] Gemini cache stats (non-streaming)", {
-              model: modelName,
-              cacheHitRate: cacheStats.cacheHitRate.toFixed(2) + "%",
-              costSavingsPercent: cacheStats.costSavingsPercent.toFixed(2) + "%",
-              totalInputTokens: cacheStats.totalInputTokens,
-              cacheReadTokens: cacheStats.cacheReadTokens
-            });
-          }
-        }
         await callback(r.candidates?.map((candidate) => candidate.content?.parts?.[0]?.text) ?? [], {
           inputTokens: r.usageMetadata?.promptTokenCount ?? 0,
           outputTokens: r.usageMetadata?.candidatesTokenCount ?? 0,
-          toolsUsed,
-          cacheStats
+          toolsUsed
         });
       } catch (error) {
         if (options.abortSignal?.aborted || error.message.includes("aborted")) {
@@ -6196,12 +5913,12 @@ var GPT5_MODELS = [
 var GPT5_1_MODELS = [ChatModels.GPT5_1, ChatModels.GPT5_1_CHAT_LATEST];
 var GPT5_2_MODELS = [ChatModels.GPT5_2, ChatModels.GPT5_2_CHAT_LATEST];
 var effortMap = {
-  simple: "minimal",
+  simple: "low",
   contextual: "low",
   complex: "medium"
 };
 var effortMap_GPT5_1_2 = {
-  simple: "none",
+  simple: "low",
   contextual: "low",
   complex: "medium"
 };
@@ -6851,20 +6568,20 @@ var OpenAIBackend = class {
         stream: true,
         ...options.maxTokens && { max_completion_tokens: options.maxTokens }
       });
-      let reasoningEffort;
-      if (options.reasoningEffort) {
-        reasoningEffort = options.reasoningEffort;
-        this.logger.debug(`Using explicit reasoning effort: ${reasoningEffort}`);
-      } else if (options.complexity && effortMap[options.complexity]) {
-        reasoningEffort = isGPT5_1Model || isGPT5_2Model ? effortMap_GPT5_1_2[options.complexity] : effortMap[options.complexity];
-        this.logger.debug(`Auto-classified reasoning effort from complexity '${options.complexity}': ${reasoningEffort}`);
-      }
-      if (reasoningEffort) {
-        Object.assign(parameters, {
-          reasoning: {
-            effort: reasoningEffort
-          }
-        });
+      if (REASONING_SUPPORTED_MODELS.has(model)) {
+        let reasoningEffort;
+        if (options.reasoningEffort) {
+          reasoningEffort = options.reasoningEffort;
+          this.logger.debug(`Using explicit reasoning effort: ${reasoningEffort}`);
+        } else if (options.complexity && effortMap[options.complexity]) {
+          reasoningEffort = isGPT5_1Model || isGPT5_2Model ? effortMap_GPT5_1_2[options.complexity] : effortMap[options.complexity];
+          this.logger.debug(`Auto-classified reasoning effort from complexity '${options.complexity}': ${reasoningEffort}`);
+        }
+        if (reasoningEffort) {
+          Object.assign(parameters, {
+            reasoning_effort: reasoningEffort
+          });
+        }
       }
     } else {
       const useStreaming = options.stream && (!options.n || options.n === 1);
@@ -6982,26 +6699,10 @@ var OpenAIBackend = class {
           streamedText[c.index] = c.message.content || "";
         }
       }
-      const cacheStrategy2 = options.cacheStrategy;
-      let cacheStats2;
-      if (cacheStrategy2?.enableCaching && response.usage) {
-        const adapter = getCachingAdapter(ModelBackend.OpenAI);
-        cacheStats2 = adapter.extractCacheStats(response, model);
-        if (cacheStats2) {
-          this.logger.info("[PromptCache] OpenAI cache stats (non-streaming)", {
-            model,
-            cacheHitRate: cacheStats2.cacheHitRate.toFixed(2) + "%",
-            costSavingsPercent: cacheStats2.costSavingsPercent.toFixed(2) + "%",
-            totalInputTokens: cacheStats2.totalInputTokens,
-            cacheReadTokens: cacheStats2.cacheReadTokens
-          });
-        }
-      }
       const completionInfo = {
         inputTokens: response.usage?.prompt_tokens || 0,
         outputTokens: response.usage?.completion_tokens || 0,
-        toolsUsed: toolsUsed.length > 0 ? toolsUsed : void 0,
-        cacheStats: cacheStats2
+        toolsUsed: toolsUsed.length > 0 ? toolsUsed : void 0
       };
       await callback(streamedText, completionInfo);
       return;
@@ -7032,33 +6733,6 @@ var OpenAIBackend = class {
         toolsUsed: toolsUsed.length > 0 ? toolsUsed : void 0
       });
     }
-    const cacheStrategy = options.cacheStrategy;
-    let cacheStats;
-    if (cacheStrategy?.enableCaching && inputTokens > 0) {
-      const adapter = getCachingAdapter(ModelBackend.OpenAI);
-      const mockResponse = {
-        usage: {
-          prompt_tokens: inputTokens,
-          completion_tokens: outputTokens,
-          prompt_tokens_details: {
-            // OpenAI provides cached token count in prompt_tokens_details
-            // We don't have direct access to it in streaming, so this will be 0
-            // unless the SDK exposes it in chunk.usage
-            cached_tokens: 0
-          }
-        }
-      };
-      cacheStats = adapter.extractCacheStats(mockResponse, model);
-      if (cacheStats && cacheStats.cacheReadTokens > 0) {
-        this.logger.info("[PromptCache] OpenAI cache stats (streaming)", {
-          model,
-          cacheHitRate: cacheStats.cacheHitRate.toFixed(2) + "%",
-          costSavingsPercent: cacheStats.costSavingsPercent.toFixed(2) + "%",
-          totalInputTokens: cacheStats.totalInputTokens,
-          cacheReadTokens: cacheStats.cacheReadTokens
-        });
-      }
-    }
     if (!isO1Model && func.length > 0) {
       for (const tool of func) {
         if (tool.name && tool.parameters) {
@@ -7083,8 +6757,7 @@ var OpenAIBackend = class {
                 await callback(results, {
                   inputTokens,
                   outputTokens,
-                  toolsUsed: toolsUsed.length > 0 ? toolsUsed : void 0,
-                  cacheStats
+                  toolsUsed: toolsUsed.length > 0 ? toolsUsed : void 0
                 });
               });
               this.pushToolMessages(messages, {
@@ -7137,8 +6810,7 @@ var OpenAIBackend = class {
         await callback([null], {
           inputTokens,
           outputTokens,
-          toolsUsed: toolsUsed.length > 0 ? toolsUsed : void 0,
-          cacheStats
+          toolsUsed: toolsUsed.length > 0 ? toolsUsed : void 0
         });
       }
     }
@@ -7461,21 +7133,8 @@ var XAIBackend = class {
     }
     const supportsThinking = await this.modelSupportsThinking(model);
     const thinkingEnabled = supportsThinking && options.thinking?.enabled !== false;
-    const cacheStrategy = options.cacheStrategy;
-    let headers = {};
-    if (cacheStrategy?.enableCaching) {
-      const adapter = getCachingAdapter(ModelBackend.XAI);
-      headers = adapter.getHeaders?.(cacheStrategy) || {};
-      if (headers["x-grok-conv-id"]) {
-        this.logger.info("[PromptCache] xAI caching enabled with conversation ID", {
-          model,
-          conversationId: headers["x-grok-conv-id"]
-        });
-      }
-    }
     const response = await this._api.chat.completions.create(parameters, {
-      signal: options.abortSignal,
-      headers
+      signal: options.abortSignal
     });
     let inputTokens = 0;
     let outputTokens = 0;
@@ -7584,25 +7243,10 @@ var XAIBackend = class {
           streamedText[c.index] = c.message.content || "";
         }
       }
-      let cacheStats2;
-      if (cacheStrategy?.enableCaching && response.usage) {
-        const adapter = getCachingAdapter(ModelBackend.XAI);
-        cacheStats2 = adapter.extractCacheStats(response, model);
-        if (cacheStats2 && cacheStats2.cacheReadTokens > 0) {
-          this.logger.info("[PromptCache] xAI cache stats (non-streaming)", {
-            model,
-            cacheHitRate: cacheStats2.cacheHitRate.toFixed(2) + "%",
-            costSavingsPercent: cacheStats2.costSavingsPercent.toFixed(2) + "%",
-            totalInputTokens: cacheStats2.totalInputTokens,
-            cacheReadTokens: cacheStats2.cacheReadTokens
-          });
-        }
-      }
       const completionInfo = {
         inputTokens: response.usage?.prompt_tokens || 0,
         outputTokens: response.usage?.completion_tokens || 0,
-        toolsUsed: toolsUsed.length > 0 ? toolsUsed : void 0,
-        cacheStats: cacheStats2
+        toolsUsed: toolsUsed.length > 0 ? toolsUsed : void 0
       };
       await callback(streamedText, completionInfo);
       return;
@@ -7647,28 +7291,6 @@ var XAIBackend = class {
         toolsUsed: toolsUsed.length > 0 ? toolsUsed : void 0
       });
     }
-    let cacheStats;
-    if (cacheStrategy?.enableCaching && inputTokens > 0) {
-      const adapter = getCachingAdapter(ModelBackend.XAI);
-      const mockResponse = {
-        usage: {
-          prompt_tokens: inputTokens,
-          completion_tokens: outputTokens,
-          cached_prompt_tokens: 0
-          // xAI provides this in the response
-        }
-      };
-      cacheStats = adapter.extractCacheStats(mockResponse, model);
-      if (cacheStats && cacheStats.cacheReadTokens > 0) {
-        this.logger.info("[PromptCache] xAI cache stats (streaming)", {
-          model,
-          cacheHitRate: cacheStats.cacheHitRate.toFixed(2) + "%",
-          costSavingsPercent: cacheStats.costSavingsPercent.toFixed(2) + "%",
-          totalInputTokens: cacheStats.totalInputTokens,
-          cacheReadTokens: cacheStats.cacheReadTokens
-        });
-      }
-    }
     if (func.length > 0) {
       for await (const tool of func) {
         const { name, parameters: parameters2 } = tool;

package/dist/{chunk-RHR4QRAI.js → chunk-65DLNQY2.js}

@@ -6,12 +6,12 @@ import {
   getSettingsByNames,
   obfuscateApiKey,
   secureParameters
-} from "./chunk-7FGWZUYS.js";
+} from "./chunk-5HA4JEOV.js";
 import {
   ApiKeyType,
   MementoTier,
   isSupportedEmbeddingModel
-} from "./chunk-LQRI6AKT.js";
+} from "./chunk-C3K7SH2V.js";
 
 // ../../b4m-core/packages/services/dist/src/apiKeyService/get.js
 import { z } from "zod";

package/dist/{chunk-ZFFJO4JD.js → chunk-B7Z2QQMD.js}

@@ -7,11 +7,11 @@ import {
   getSettingsMap,
   getSettingsValue,
   secureParameters
-} from "./chunk-7FGWZUYS.js";
+} from "./chunk-5HA4JEOV.js";
 import {
   KnowledgeType,
   SupportedFabFileMimeTypes
-} from "./chunk-LQRI6AKT.js";
+} from "./chunk-C3K7SH2V.js";
 
 // ../../b4m-core/packages/services/dist/src/fabFileService/create.js
 import { z } from "zod";

package/dist/{chunk-Q64OUQDG.js → chunk-BYLWQZ5S.js}

@@ -2,7 +2,7 @@
 import {
   BadRequestError,
   secureParameters
-} from "./chunk-7FGWZUYS.js";
+} from "./chunk-5HA4JEOV.js";
 import {
   CompletionApiUsageTransaction,
   GenericCreditDeductTransaction,
@@ -12,7 +12,7 @@ import {
   TextGenerationUsageTransaction,
   TransferCreditTransaction,
   VideoGenerationUsageTransaction
-} from "./chunk-LQRI6AKT.js";
+} from "./chunk-C3K7SH2V.js";
 
 // ../../b4m-core/packages/services/dist/src/creditService/subtractCredits.js
 import { z } from "zod";

package/dist/{chunk-LQRI6AKT.js → chunk-C3K7SH2V.js}

@@ -171,6 +171,20 @@ var ChatModels;
 })(ChatModels || (ChatModels = {}));
 var CHAT_MODELS = Object.values(ChatModels);
 var supportedChatModels = z.nativeEnum(ChatModels);
+var REASONING_SUPPORTED_MODELS = /* @__PURE__ */ new Set([
+  ChatModels.O1,
+  ChatModels.O3_MINI,
+  ChatModels.O3,
+  ChatModels.O4_MINI,
+  ChatModels.GPT5,
+  ChatModels.GPT5_MINI,
+  ChatModels.GPT5_NANO,
+  ChatModels.GPT5_CHAT_LATEST,
+  ChatModels.GPT5_1,
+  ChatModels.GPT5_1_CHAT_LATEST,
+  ChatModels.GPT5_2,
+  ChatModels.GPT5_2_CHAT_LATEST
+]);
 var SpeechToTextModels;
 (function(SpeechToTextModels2) {
   SpeechToTextModels2["WHISPER_1"] = "whisper-1";
@@ -8595,6 +8609,7 @@ export {
   ChatModels,
   CHAT_MODELS,
   supportedChatModels,
+  REASONING_SUPPORTED_MODELS,
   SpeechToTextModels,
   SPEECH_TO_TEXT_MODELS,
   supportedSpeechToTextModels,

package/dist/{create-EB4BK7CX.js → create-IDGWY5LO.js}

@@ -2,9 +2,9 @@
 import {
   createFabFile,
   createFabFileSchema
-} from "./chunk-ZFFJO4JD.js";
-import "./chunk-7FGWZUYS.js";
-import "./chunk-LQRI6AKT.js";
+} from "./chunk-B7Z2QQMD.js";
+import "./chunk-5HA4JEOV.js";
+import "./chunk-C3K7SH2V.js";
 import "./chunk-OCYRD7D6.js";
 export {
   createFabFile,

package/dist/index.js

@@ -5,7 +5,7 @@ import {
   getEffectiveApiKey,
   getOpenWeatherKey,
   getSerperKey
-} from "./chunk-RHR4QRAI.js";
+} from "./chunk-65DLNQY2.js";
 import "./chunk-RUI6HNLO.js";
 import {
   ConfigStore,
@@ -15,8 +15,8 @@ import {
   selectActiveBackgroundAgents,
   useCliStore
 } from "./chunk-TVW4ZESU.js";
-import "./chunk-Q64OUQDG.js";
-import "./chunk-ZFFJO4JD.js";
+import "./chunk-BYLWQZ5S.js";
+import "./chunk-B7Z2QQMD.js";
 import {
   BFLImageService,
   BaseStorage,
@@ -28,7 +28,7 @@ import {
   OpenAIBackend,
   OpenAIImageService,
   XAIImageService
-} from "./chunk-7FGWZUYS.js";
+} from "./chunk-5HA4JEOV.js";
 import {
   AiEvents,
   ApiKeyEvents,
@@ -84,7 +84,7 @@ import {
   XAI_IMAGE_MODELS,
   b4mLLMTools,
   getMcpProviderMetadata
-} from "./chunk-LQRI6AKT.js";
+} from "./chunk-C3K7SH2V.js";
 import {
   Logger
 } from "./chunk-OCYRD7D6.js";
@@ -316,9 +316,23 @@ function CommandAutocomplete({ commands, selectedIndex }) {
   if (commands.length === 0) {
     return /* @__PURE__ */ React3.createElement(Box2, { marginLeft: 2, marginTop: 1 }, /* @__PURE__ */ React3.createElement(Text3, { dimColor: true }, "No matching commands"));
   }
-  return /* @__PURE__ */ React3.createElement(Box2, { flexDirection: "column", marginLeft: 2, marginTop: 1 }, /* @__PURE__ */ React3.createElement(Box2, { marginBottom: 1 }, /* @__PURE__ */ React3.createElement(Text3, { dimColor: true }, commands.length === 1 ? "1 command" : `${commands.length} commands`, " \u2022 Use \u2191\u2193 to navigate, Enter to select")), commands.map((cmd, index) => {
+  const VIEWPORT_SIZE = 6;
+  const totalCommands = commands.length;
+  let startIndex = 0;
+  let endIndex = Math.min(VIEWPORT_SIZE, totalCommands);
+  if (totalCommands > VIEWPORT_SIZE) {
+    const halfViewport = Math.floor(VIEWPORT_SIZE / 2);
+    startIndex = Math.max(0, selectedIndex - halfViewport);
+    endIndex = Math.min(totalCommands, startIndex + VIEWPORT_SIZE);
+    if (endIndex === totalCommands) {
+      startIndex = Math.max(0, totalCommands - VIEWPORT_SIZE);
+    }
+  }
+  const visibleCommands = commands.slice(startIndex, endIndex);
+  return /* @__PURE__ */ React3.createElement(Box2, { flexDirection: "column", marginLeft: 2, marginTop: 1 }, /* @__PURE__ */ React3.createElement(Box2, { marginBottom: 1 }, /* @__PURE__ */ React3.createElement(Text3, { dimColor: true }, totalCommands === 1 ? "1 command" : `${totalCommands} commands`, totalCommands > VIEWPORT_SIZE && ` (${selectedIndex + 1}/${totalCommands})`, " \u2022 Use \u2191\u2193 to navigate, Enter to select")), visibleCommands.map((cmd, viewportIndex) => {
+    const actualIndex = startIndex + viewportIndex;
     const args = cmd.args ? ` ${cmd.args}` : "";
-    const isSelected = index === selectedIndex;
+    const isSelected = actualIndex === selectedIndex;
     const sourceIcon = cmd.source === "global" ? "\u{1F3E0} " : cmd.source === "project" ? "\u{1F4C1} " : cmd.source === "built-in" ? "\u{1F527} " : "";
     return /* @__PURE__ */ React3.createElement(Box2, { key: cmd.name, marginLeft: 1 }, /* @__PURE__ */ React3.createElement(Text3, { color: isSelected ? "cyan" : void 0, bold: isSelected }, isSelected ? "\u25B8 " : "  ", sourceIcon, "/", cmd.name, args, " - ", cmd.description));
   }));
@@ -14636,7 +14650,7 @@ import { isAxiosError as isAxiosError2 } from "axios";
 // package.json
 var package_default = {
   name: "@bike4mind/cli",
-  version: "0.2.30-feat-universal-prompt-caching.19181+42051b4f5",
+  version: "0.2.30-feat-cli-scrollable-command-autocomplete.19203+b74387251",
   type: "module",
   description: "Interactive CLI tool for Bike4Mind with ReAct agents",
   license: "UNLICENSED",
@@ -14707,13 +14721,10 @@ var package_default = {
     diff: "^8.0.2",
     dotenv: "^16.3.1",
     "eventsource-parser": "^3.0.6",
-    fdir: "^6.5.0",
     "file-type": "^18.7.0",
     "fuse.js": "^7.1.0",
-    fzf: "^0.5.2",
     glob: "^13.0.0",
     "gray-matter": "^4.0.3",
-    ignore: "^7.0.5",
     ink: "^6.5.1",
     "ink-select-input": "^6.2.0",
     "ink-spinner": "^5.0.0",
@@ -14750,10 +14761,10 @@ var package_default = {
   },
   devDependencies: {
     "@bike4mind/agents": "0.1.0",
-    "@bike4mind/common": "2.51.1-feat-universal-prompt-caching.19181+42051b4f5",
-    "@bike4mind/mcp": "1.30.1-feat-universal-prompt-caching.19181+42051b4f5",
-    "@bike4mind/services": "2.49.1-feat-universal-prompt-caching.19181+42051b4f5",
-    "@bike4mind/utils": "2.6.1-feat-universal-prompt-caching.19181+42051b4f5",
+    "@bike4mind/common": "2.51.1-feat-cli-scrollable-command-autocomplete.19203+b74387251",
+    "@bike4mind/mcp": "1.30.1-feat-cli-scrollable-command-autocomplete.19203+b74387251",
+    "@bike4mind/services": "2.49.1-feat-cli-scrollable-command-autocomplete.19203+b74387251",
+    "@bike4mind/utils": "2.6.1-feat-cli-scrollable-command-autocomplete.19203+b74387251",
     "@types/better-sqlite3": "^7.6.13",
     "@types/diff": "^5.0.9",
     "@types/jsonwebtoken": "^9.0.4",
@@ -14762,6 +14773,9 @@ var package_default = {
     "@types/react": "^19.2.7",
     "@types/uuid": "^9.0.7",
     "@types/yargs": "^17.0.32",
+    fdir: "^6.5.0",
+    fzf: "^0.5.2",
+    ignore: "^7.0.5",
     "ink-testing-library": "^4.0.0",
     tsup: "^8.5.1",
     tsx: "^4.21.0",
@@ -14771,7 +14785,7 @@ var package_default = {
   optionalDependencies: {
     "@vscode/ripgrep": "^1.17.0"
   },
-  gitHead: "42051b4f55164900b7d8377a46c89d4c63c4f334"
+  gitHead: "b74387251e07e79e94b8720e6d4b1c8e16bde324"
 };
 
 // src/config/constants.ts

package/dist/{llmMarkdownGenerator-L4FJDAYW.js → llmMarkdownGenerator-BILRFSX6.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import {
   CurationArtifactType
-} from "./chunk-LQRI6AKT.js";
+} from "./chunk-C3K7SH2V.js";
 
 // ../../b4m-core/packages/services/dist/src/notebookCurationService/llmMarkdownGenerator.js
 var DEFAULT_OPTIONS = {

package/dist/{markdownGenerator-TWB6RRAR.js → markdownGenerator-AI3EZ4Z3.js}

@@ -1,7 +1,7 @@
 #!/usr/bin/env node
 import {
   CurationArtifactType
-} from "./chunk-LQRI6AKT.js";
+} from "./chunk-C3K7SH2V.js";
 
 // ../../b4m-core/packages/services/dist/src/notebookCurationService/markdownGenerator.js
 var DEFAULT_OPTIONS = {

package/dist/{mementoService-ICKCCKSL.js → mementoService-34S5SRX4.js}

@@ -2,9 +2,9 @@
 import {
   findMostSimilarMemento,
   getRelevantMementos
-} from "./chunk-RHR4QRAI.js";
-import "./chunk-7FGWZUYS.js";
-import "./chunk-LQRI6AKT.js";
+} from "./chunk-65DLNQY2.js";
+import "./chunk-5HA4JEOV.js";
+import "./chunk-C3K7SH2V.js";
 import "./chunk-OCYRD7D6.js";
 export {
   findMostSimilarMemento,

package/dist/{src-7CHMPY4J.js → src-MDS22ESD.js}

@@ -134,12 +134,12 @@ import {
   validateMermaidSyntax,
   warmUpSettingsCache,
   withRetry
-} from "./chunk-7FGWZUYS.js";
+} from "./chunk-5HA4JEOV.js";
 import {
   buildRateLimitLogEntry,
   isNearLimit,
   parseRateLimitHeaders
-} from "./chunk-LQRI6AKT.js";
+} from "./chunk-C3K7SH2V.js";
 import {
   Logger,
   NotificationDeduplicator,

package/dist/{src-3EWLI2XV.js → src-QX72HMCT.js}

@@ -232,6 +232,7 @@ import {
   QuestStatusSchema,
   REASONING_EFFORT_DESCRIPTIONS,
   REASONING_EFFORT_LABELS,
+  REASONING_SUPPORTED_MODELS,
   RESTRICTION_OPERATIONS,
   RESTRICTION_SUBJECT_TYPES,
   RapidReplyFallbackBehaviors,
@@ -418,7 +419,7 @@ import {
   validateReactArtifactV2,
   validateSvgArtifactV2,
   wikiMarkupToAdf
-} from "./chunk-LQRI6AKT.js";
+} from "./chunk-C3K7SH2V.js";
 export {
   ALL_IMAGE_MODELS,
   ALL_IMAGE_SIZES,
@@ -653,6 +654,7 @@ export {
   QuestStatusSchema,
   REASONING_EFFORT_DESCRIPTIONS,
   REASONING_EFFORT_LABELS,
+  REASONING_SUPPORTED_MODELS,
   RESTRICTION_OPERATIONS,
   RESTRICTION_SUBJECT_TYPES,
   RapidReplyFallbackBehaviors,

package/dist/{subtractCredits-G6ZJLDCV.js → subtractCredits-6CK3FPNG.js}

@@ -2,9 +2,9 @@
 import {
   SubtractCreditsSchema,
   subtractCredits
-} from "./chunk-Q64OUQDG.js";
-import "./chunk-7FGWZUYS.js";
-import "./chunk-LQRI6AKT.js";
+} from "./chunk-BYLWQZ5S.js";
+import "./chunk-5HA4JEOV.js";
+import "./chunk-C3K7SH2V.js";
 import "./chunk-OCYRD7D6.js";
 export {
   SubtractCreditsSchema,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@bike4mind/cli",
-  "version": "0.2.30-feat-universal-prompt-caching.19181+42051b4f5",
+  "version": "0.2.30-feat-cli-scrollable-command-autocomplete.19203+b74387251",
   "type": "module",
   "description": "Interactive CLI tool for Bike4Mind with ReAct agents",
   "license": "UNLICENSED",
@@ -71,13 +71,10 @@
     "diff": "^8.0.2",
     "dotenv": "^16.3.1",
     "eventsource-parser": "^3.0.6",
-    "fdir": "^6.5.0",
     "file-type": "^18.7.0",
     "fuse.js": "^7.1.0",
-    "fzf": "^0.5.2",
     "glob": "^13.0.0",
     "gray-matter": "^4.0.3",
-    "ignore": "^7.0.5",
     "ink": "^6.5.1",
     "ink-select-input": "^6.2.0",
     "ink-spinner": "^5.0.0",
@@ -114,10 +111,10 @@
   },
   "devDependencies": {
     "@bike4mind/agents": "0.1.0",
-    "@bike4mind/common": "2.51.1-feat-universal-prompt-caching.19181+42051b4f5",
-    "@bike4mind/mcp": "1.30.1-feat-universal-prompt-caching.19181+42051b4f5",
-    "@bike4mind/services": "2.49.1-feat-universal-prompt-caching.19181+42051b4f5",
-    "@bike4mind/utils": "2.6.1-feat-universal-prompt-caching.19181+42051b4f5",
+    "@bike4mind/common": "2.51.1-feat-cli-scrollable-command-autocomplete.19203+b74387251",
+    "@bike4mind/mcp": "1.30.1-feat-cli-scrollable-command-autocomplete.19203+b74387251",
+    "@bike4mind/services": "2.49.1-feat-cli-scrollable-command-autocomplete.19203+b74387251",
+    "@bike4mind/utils": "2.6.1-feat-cli-scrollable-command-autocomplete.19203+b74387251",
     "@types/better-sqlite3": "^7.6.13",
     "@types/diff": "^5.0.9",
     "@types/jsonwebtoken": "^9.0.4",
@@ -126,6 +123,9 @@
     "@types/react": "^19.2.7",
     "@types/uuid": "^9.0.7",
     "@types/yargs": "^17.0.32",
+    "fdir": "^6.5.0",
+    "fzf": "^0.5.2",
+    "ignore": "^7.0.5",
     "ink-testing-library": "^4.0.0",
     "tsup": "^8.5.1",
     "tsx": "^4.21.0",
@@ -135,5 +135,5 @@
   "optionalDependencies": {
     "@vscode/ripgrep": "^1.17.0"
   },
-  "gitHead": "42051b4f55164900b7d8377a46c89d4c63c4f334"
+  "gitHead": "b74387251e07e79e94b8720e6d4b1c8e16bde324"
 }