Merge origin/master (v0.5.75) into gitea/new_feature

Resolve conflicts:
- package.json / cli/package.json: take 0.5.75
- .gitignore: union both sides (upstream 9router-*/temp files + local state dirs)
- CHANGELOG.md: keep both blocks, v0.5.75 above v0.5.70
- nonStreamingHandler.js: merge imports (unwrapClineEnvelope +
  tokensForDetail/shouldPersistRequestDetail); drop dead appendRequestLog
- providers/[id]/page.js: union useState blocks (compatible-model states
  + importingClineModels)

Co-authored-by: CommandCodeBot <noreply@commandcode.ai>

open-sse/translator/request/claude-to-kiro.js

@@ -242,9 +242,9 @@ export function claudeToKiroRequest(model, body, stream, credentials) {
     ? (credentials?.providerSpecificData?.profileArn || "")
     : (credentials?.providerSpecificData?.profileArn || resolveDefaultProfileArn(authMethod));
 
-  // Kiro CLI/KAS sends system prompt as top-level `systemPrompt`. Keep a
-  // content fallback too because the CodeWhisperer surface does not always
-  // enforce top-level systemPrompt for direct calls.
+  // The system prompt travels inside the first user turn's content (contentPrefix):
+  // the CodeWhisperer surface rejects a top-level `systemPrompt` with
+  // 400 REQUEST_BODY_INVALID, so the value below is only a replay cache key.
   const timestamp = new Date().toISOString();
   const systemPromptParts = [];
   if (thinkingBudget !== null && !usesNativeGptEffort) {
@@ -316,14 +316,11 @@ export function claudeToKiroRequest(model, body, stream, credentials) {
     conversationState: {
       chatTriggerType: "MANUAL",
       conversationId,
-      agentContinuationId: continuationId,
-      agentTaskType: "vibe",
       currentMessage: {
         userInputMessage,
       },
       history: canonical.history,
     },
-    agentMode: "vibe",
   };
 
   if (profileArn) payload.profileArn = profileArn;

open-sse/translator/request/claude-to-openai.js

@@ -142,6 +142,13 @@ function systemReminderText(content) {
 
 // Convert single Claude message - returns single message or array of messages
 function convertClaudeMessage(msg) {
+  // Some clients send content as a single block object; normalize to the
+  // one-element array every branch below (the system-reminder fold included)
+  // expects. Must run BEFORE the role branch: systemReminderText only reads
+  // arrays and strings, so a bare-object system turn was dropped outright.
+  if (msg.content && typeof msg.content === "object" && !Array.isArray(msg.content)) {
+    msg.content = [msg.content];
+  }
   // Mid-conversation system message -> user (per Anthropic placement rules)
   if (msg.role === ROLE.SYSTEM) {
     const text = systemReminderText(msg.content);

open-sse/translator/request/openai-to-gemini.js

@@ -15,7 +15,8 @@ import {
   generateRequestId,
   generateSessionId,
   generateProjectId,
-  cleanJSONSchemaForAntigravity
+  cleanJSONSchemaForAntigravity,
+  normalizeGeminiContents
 } from "../formats/gemini.js";
 import { deriveSessionId, toNumericSessionId } from "../../utils/sessionManager.js";
 import { ROLE, GEMINI_ROLE, OPENAI_BLOCK, CLAUDE_BLOCK } from "../schema/index.js";
@@ -35,17 +36,6 @@ function sanitizeGeminiFunctionName(name) {
   return sanitized.substring(0, 64);
 }
 
-function normalizeGeminiContents(contents) {
-  const out = [];
-  for (const c of contents || []) {
-    if (!c?.role || !Array.isArray(c.parts) || c.parts.length === 0) continue;
-    const last = out.at(-1);
-    if (last?.role === c.role) last.parts.push(...c.parts);
-    else out.push({ ...c, parts: [...c.parts] });
-  }
-  return out;
-}
-
 // Core: Convert OpenAI request to Gemini format (base for all variants)
 function openaiToGeminiBase(model, body, stream, signature = DEFAULT_THINKING_AG_SIGNATURE, sessionId = null) {
   const result = {
@@ -163,12 +153,14 @@ function openaiToGeminiBase(model, body, stream, signature = DEFAULT_THINKING_AG
           }
 
           // Check if there are actual tool responses in the next messages
-          const hasActualResponses = toolCallIds.some(fid => toolResponses[fid]);
+          const isIntermediate = i < body.messages.length - 1;
+          const hasActualResponses = toolCallIds.some(fid => toolResponses[fid] !== undefined);
 
-          if (hasActualResponses) {
+          if (hasActualResponses || isIntermediate) {
             const toolParts = [];
             for (const fid of toolCallIds) {
-              if (!toolResponses[fid]) continue;
+              let resp = toolResponses[fid];
+              if (resp === undefined) resp = "";
 
               let name = tcID2Name[fid];
               if (!name) {
@@ -180,7 +172,6 @@ function openaiToGeminiBase(model, body, stream, signature = DEFAULT_THINKING_AG
                 }
               }
 
-              let resp = toolResponses[fid];
               let parsedResp = tryParseJSON(resp);
               if (parsedResp === null) {
                 parsedResp = { result: resp };

open-sse/translator/request/openai-to-kiro.js

@@ -340,9 +340,9 @@ export function openaiToKiroRequest(model, body, stream, credentials) {
 
   const timestamp = new Date().toISOString();
 
-  // Kiro CLI/KAS sends these as top-level systemPrompt. Keep a content fallback
-  // too because the CodeWhisperer surface does not always enforce top-level
-  // systemPrompt for direct calls.
+  // The system prompt travels inside the first user turn's content (contentPrefix):
+  // the CodeWhisperer surface rejects a top-level `systemPrompt` with
+  // 400 REQUEST_BODY_INVALID, so the value below is only a replay cache key.
   const systemPromptParts = [];
   if (thinkingBudget !== null && !usesNativeGptEffort) {
     systemPromptParts.push(buildThinkingSystemPrefix(thinkingBudget));
@@ -397,8 +397,6 @@ export function openaiToKiroRequest(model, body, stream, credentials) {
     conversationState: {
       chatTriggerType: "MANUAL",
       conversationId,
-      agentContinuationId: continuationId,
-      agentTaskType: "vibe",
       currentMessage: {
         userInputMessage: {
           content: replayCurrent.content || "",
@@ -414,7 +412,6 @@ export function openaiToKiroRequest(model, body, stream, credentials) {
       },
       history: canonical.history
     },
-    agentMode: "vibe",
   };
 
   if (profileArn) {