pybao-cli 1.5.16 → 1.5.18

package/dist/{customCommands-6TIJA6HB.js → customCommands-5FLRWSSP.js}

@@ -8,11 +8,11 @@ import {
   parseFrontmatter,
   reloadCustomCommands,
   resolveFileReferences
-} from "./chunk-5NEN4RZF.js";
+} from "./chunk-PMFFKAFV.js";
 import "./chunk-BJSWTHRM.js";
-import "./chunk-GF2MQGGQ.js";
-import "./chunk-TKUXXEIH.js";
-import "./chunk-ZVKBWRCR.js";
+import "./chunk-UEU42RLM.js";
+import "./chunk-HE3VVGNG.js";
+import "./chunk-TVQIMRQR.js";
 import "./chunk-I3J4JYES.js";
 export {
   executeBashCommands,

package/dist/{env-7RX7XQ4A.js → env-IONEA5NZ.js}

@@ -15,8 +15,8 @@ import {
   isContextScanDisabled,
   resolveXdgCachePath,
   resolveXdgDataPath
-} from "./chunk-TKUXXEIH.js";
-import "./chunk-ZVKBWRCR.js";
+} from "./chunk-HE3VVGNG.js";
+import "./chunk-TVQIMRQR.js";
 import "./chunk-I3J4JYES.js";
 export {
   GLOBAL_CONFIG_FILE,

package/dist/{file-LJN7Y7PR.js → file-SYRHIZ77.js}

@@ -18,10 +18,10 @@ import {
   readFileSafe,
   readTextContent,
   writeTextContent
-} from "./chunk-7TRQANX3.js";
-import "./chunk-6UUMMCEA.js";
-import "./chunk-TKUXXEIH.js";
-import "./chunk-ZVKBWRCR.js";
+} from "./chunk-Q77WSFTD.js";
+import "./chunk-OWVJDCD3.js";
+import "./chunk-HE3VVGNG.js";
+import "./chunk-TVQIMRQR.js";
 import "./chunk-I3J4JYES.js";
 export {
   addLineNumbers,

package/dist/index.js

@@ -2,7 +2,7 @@ import { createRequire as __pybCreateRequire } from "node:module";
 const require = __pybCreateRequire(import.meta.url);
 import {
   MACRO
-} from "./chunk-ZVKBWRCR.js";
+} from "./chunk-TVQIMRQR.js";
 import "./chunk-I3J4JYES.js";
 
 // src/entrypoints/index.ts
@@ -28,7 +28,7 @@ Common options:
   process.exit(0);
 }
 if (hasFlag("--acp")) {
-  await import("./acp-OQU3Q74X.js");
+  await import("./acp-E7PSBZVL.js");
 } else {
-  await import("./cli-BQVVZA5X.js");
+  await import("./cli-NAGNW2KG.js");
 }

package/dist/{llm-NAYVED3B.js → llm-VJSYR35Z.js}

@@ -2,7 +2,7 @@ import { createRequire as __pybCreateRequire } from "node:module";
 const require = __pybCreateRequire(import.meta.url);
 import {
   getToolDescription
-} from "./chunk-5ZAXW6YC.js";
+} from "./chunk-BKCAVW2G.js";
 import {
   formatSystemPromptWithContext,
   generatePybContext,
@@ -10,29 +10,29 @@ import {
   getReasoningEffort,
   models_default,
   refreshPybContext
-} from "./chunk-LE4MSECX.js";
-import "./chunk-ZGQA3IBT.js";
-import "./chunk-ZVVBLU3R.js";
-import "./chunk-DXLMPRXX.js";
-import "./chunk-GZBASJQM.js";
-import "./chunk-74PPXIKF.js";
-import "./chunk-MMTMYYK4.js";
+} from "./chunk-SI3S7CU7.js";
+import "./chunk-ZLU7DVAO.js";
+import "./chunk-SLHGOQLQ.js";
+import "./chunk-W5XUAORH.js";
+import "./chunk-ZQFDJJAQ.js";
+import "./chunk-TOWF6PJA.js";
+import "./chunk-JVU5Z6OG.js";
 import "./chunk-F4AXICO7.js";
 import "./chunk-5P7HBXTD.js";
-import "./chunk-OTEQDHXR.js";
-import "./chunk-ZHCJRYGZ.js";
+import "./chunk-VNHCWERF.js";
+import "./chunk-EMJK4OX7.js";
 import {
   getCompletionWithProfile,
   getGPT5CompletionWithProfile
-} from "./chunk-2UIC6KUN.js";
+} from "./chunk-5W5QFMDG.js";
 import "./chunk-XKYHFZEC.js";
-import "./chunk-WCNJUZPB.js";
-import "./chunk-2JDFT27Z.js";
-import "./chunk-GDUBW6S3.js";
-import "./chunk-XC7HSASO.js";
+import "./chunk-6KEMNSSL.js";
+import "./chunk-LK7EWRRJ.js";
+import "./chunk-MEYAIOUR.js";
+import "./chunk-PD5MV3BL.js";
 import "./chunk-UNNVICVU.js";
-import "./chunk-2IP3S5YY.js";
-import "./chunk-OQUAEZWY.js";
+import "./chunk-X52WVHNZ.js";
+import "./chunk-MH6INOZV.js";
 import "./chunk-A3BVXXA3.js";
 import {
   API_ERROR_MESSAGE_PREFIX,
@@ -44,7 +44,7 @@ import {
   createAssistantAPIErrorMessage,
   emitTelemetryEvent,
   normalizeContentFromAPI
-} from "./chunk-EGFEUHPK.js";
+} from "./chunk-SMRXVPJO.js";
 import {
   processResponsesStream
 } from "./chunk-OMELVAJD.js";
@@ -55,27 +55,27 @@ import "./chunk-B6IMQJZM.js";
 import {
   addToTotalCost
 } from "./chunk-OUXHGDLH.js";
-import "./chunk-7TRQANX3.js";
+import "./chunk-Q77WSFTD.js";
 import "./chunk-QWIBSCDN.js";
-import "./chunk-7UPV7QTD.js";
-import "./chunk-SQDHH3JF.js";
-import "./chunk-TOLIZZNG.js";
+import "./chunk-JQEVUTXP.js";
+import "./chunk-AU3KUCWJ.js";
+import "./chunk-H6IALBZQ.js";
 import "./chunk-UZ34JEUK.js";
-import "./chunk-5NEN4RZF.js";
+import "./chunk-PMFFKAFV.js";
 import "./chunk-BJSWTHRM.js";
 import {
   USE_BEDROCK,
   USE_VERTEX,
   getModelManager,
   getVertexRegionForModel
-} from "./chunk-PJ6F6UEG.js";
-import "./chunk-VLAJHHDH.js";
-import "./chunk-6UUMMCEA.js";
-import "./chunk-IW5UFMRQ.js";
+} from "./chunk-7MP24CSX.js";
+import "./chunk-KX27QO65.js";
+import "./chunk-OWVJDCD3.js";
+import "./chunk-IYLRQYZC.js";
 import {
   getAnthropicApiKey,
   getGlobalConfig
-} from "./chunk-T2TMNQ7H.js";
+} from "./chunk-XIVGJEPA.js";
 import "./chunk-RQVLBMP7.js";
 import {
   debug,
@@ -84,16 +84,16 @@ import {
   logLLMInteraction,
   logSystemPromptConstruction,
   markPhase
-} from "./chunk-GF2MQGGQ.js";
+} from "./chunk-UEU42RLM.js";
 import {
   PRODUCT_COMMAND,
   env,
   getCwd,
   logError
-} from "./chunk-TKUXXEIH.js";
+} from "./chunk-HE3VVGNG.js";
 import {
   MACRO
-} from "./chunk-ZVKBWRCR.js";
+} from "./chunk-TVQIMRQR.js";
 import "./chunk-I3J4JYES.js";
 
 // src/services/ai/llm.ts
@@ -1359,8 +1359,11 @@ var ResponseStateManager = class {
     }, this.CLEANUP_INTERVAL);
   }
   setPreviousResponseId(conversationId, responseId) {
+    const previous = this.conversationStates.get(conversationId);
+    const nextDepth = Number.isFinite(Number(previous?.responseChainDepth)) && Number(previous?.responseChainDepth) > 0 ? Number(previous?.responseChainDepth) + 1 : 1;
     this.conversationStates.set(conversationId, {
       previousResponseId: responseId,
+      responseChainDepth: nextDepth,
       lastUpdate: Date.now()
     });
   }
@@ -1372,6 +1375,17 @@ var ResponseStateManager = class {
     }
     return void 0;
   }
+  getResponseChainDepth(conversationId) {
+    const state = this.conversationStates.get(conversationId);
+    if (!state) {
+      return 0;
+    }
+    state.lastUpdate = Date.now();
+    if (Number.isFinite(Number(state.responseChainDepth)) && Number(state.responseChainDepth) >= 0) {
+      return Number(state.responseChainDepth);
+    }
+    return state.previousResponseId ? 1 : 0;
+  }
   clearConversation(conversationId) {
     this.conversationStates.delete(conversationId);
   }
@@ -2593,15 +2607,26 @@ function getAnthropicClient(model) {
 function resetAnthropicClient() {
   anthropicClient = null;
 }
+function isCompactedBoundaryText(text) {
+  return text.includes("<conversation-summary>") || text.includes("tool.state.time.compacted");
+}
 function applyCacheControlWithLimits(systemBlocks, messageParams) {
   if (!PROMPT_CACHING_ENABLED) {
-    return { systemBlocks, messageParams };
+    return {
+      systemBlocks,
+      messageParams,
+      cached_block_count: 0,
+      cached_block_types: []
+    };
   }
   const maxCacheBlocks = 4;
   let usedCacheBlocks = 0;
+  const cachedBlockTypes = [];
   const processedSystemBlocks = systemBlocks.map((block, index) => {
-    if (usedCacheBlocks < maxCacheBlocks && block.text.length > 1e3) {
+    const compactedBoundaryBlock = isCompactedBoundaryText(block.text);
+    if (usedCacheBlocks < maxCacheBlocks && block.text.length > 1e3 && !compactedBoundaryBlock) {
       usedCacheBlocks++;
+      cachedBlockTypes.push(`system:${index}`);
       return {
         ...block,
         cache_control: { type: "ephemeral" }
@@ -2614,9 +2639,10 @@ function applyCacheControlWithLimits(systemBlocks, messageParams) {
     if (Array.isArray(message.content)) {
       const processedContent = message.content.map(
         (contentBlock, blockIndex) => {
-          const shouldCache = usedCacheBlocks < maxCacheBlocks && contentBlock.type === "text" && typeof contentBlock.text === "string" && (contentBlock.text.length > 2e3 || messageIndex === messageParams.length - 1 && blockIndex === message.content.length - 1 && contentBlock.text.length > 500);
+          const shouldCache = usedCacheBlocks < maxCacheBlocks && contentBlock.type === "text" && typeof contentBlock.text === "string" && !isCompactedBoundaryText(contentBlock.text) && (contentBlock.text.length > 2e3 || messageIndex === messageParams.length - 1 && blockIndex === message.content.length - 1 && contentBlock.text.length > 500);
           if (shouldCache) {
             usedCacheBlocks++;
+            cachedBlockTypes.push(`message:${messageIndex}:${blockIndex}`);
             return {
               ...contentBlock,
               cache_control: { type: "ephemeral" }
@@ -2633,10 +2659,14 @@ function applyCacheControlWithLimits(systemBlocks, messageParams) {
     }
     return message;
   });
-  return {
+  const payload = {
     systemBlocks: processedSystemBlocks,
-    messageParams: processedMessageParams
+    messageParams: processedMessageParams,
+    cached_block_count: usedCacheBlocks,
+    cached_block_types: cachedBlockTypes
   };
+  debug.api("CACHE_CONTROL_APPLIED", payload);
+  return payload;
 }
 function userMessageToMessageParam(message, addCache = false) {
   if (addCache) {
@@ -2734,9 +2764,11 @@ async function queryLLM(messages, systemPrompt, maxThinkingTokens, tools, signal
       toolUseContext.messageId
     );
     const previousResponseId = responseStateManager.getPreviousResponseId(conversationId);
+    const responseChainDepth = responseStateManager.getResponseChainDepth(conversationId);
     toolUseContext.responseState = {
       previousResponseId,
-      conversationId
+      conversationId,
+      responseChainDepth
     };
   }
   debug.api("MODEL_RESOLVED", {
@@ -2746,6 +2778,7 @@ async function queryLLM(messages, systemPrompt, maxThinkingTokens, tools, signal
     isPointer: ["main", "task", "compact", "quick"].includes(options.model),
     hasResponseState: !!toolUseContext?.responseState,
     conversationId: toolUseContext?.responseState?.conversationId,
+    responseChainDepth: toolUseContext?.responseState?.responseChainDepth,
     requestId: getCurrentRequest()?.id
   });
   const currentRequest = getCurrentRequest();
@@ -2909,7 +2942,21 @@ async function queryAnthropicNative(messages, systemPrompt, maxThinkingTokens, t
     )
   );
   const anthropicMessages = addCacheBreakpoints(messages);
-  const { systemBlocks: processedSystem, messageParams: processedMessages } = applyCacheControlWithLimits(system, anthropicMessages);
+  const {
+    systemBlocks: processedSystem,
+    messageParams: processedMessages,
+    cached_block_count,
+    cached_block_types
+  } = applyCacheControlWithLimits(system, anthropicMessages);
+  debug.api("CACHE_CONTROL_CHAIN", {
+    cached_block_count,
+    cached_block_types,
+    conversationId: options?.toolUseContext?.responseState?.conversationId,
+    response_chain_depth: options?.toolUseContext?.responseState?.responseChainDepth ?? 0,
+    has_previous_response_id: Boolean(
+      options?.toolUseContext?.responseState?.previousResponseId
+    )
+  });
   const startIncludingRetries = Date.now();
   logSystemPromptConstruction({
     basePrompt: systemPrompt.join("\n"),
@@ -3308,7 +3355,7 @@ async function queryOpenAI(messages, systemPrompt, maxThinkingTokens, tools, sig
         start = Date.now();
         if (adapterContext) {
           if (adapterContext.shouldUseResponses) {
-            const { callGPT5ResponsesAPI } = await import("./openai-FQAJSMPU.js");
+            const { callGPT5ResponsesAPI } = await import("./openai-UOZRW5BW.js");
             const response = await callGPT5ResponsesAPI(
               modelProfile,
               adapterContext.request,
@@ -3568,6 +3615,7 @@ export {
   MAIN_QUERY_TEMPERATURE,
   NO_CONTENT_MESSAGE,
   PROMPT_TOO_LONG_ERROR_MESSAGE,
+  applyCacheControlWithLimits,
   assistantMessageToMessageParam,
   fetchAnthropicModels,
   formatSystemPromptWithContext,