@nomad-e/bluma-cli 0.26.1 → 0.26.3

package/dist/main.js

@@ -25711,6 +25711,9 @@ function resolveReasoningEffortForRequest(params, runtimeConfig = getRuntimeConf
   }
   return effort;
 }
+function isReasoningEnabledForRequest(params, runtimeConfig = getRuntimeConfig()) {
+  return resolveReasoningEffortForRequest(params, runtimeConfig) !== void 0;
+}
 function buildChatCompletionRequestBody(params, runtimeConfig = getRuntimeConfig(), stream = false, options) {
   const tools = params.tools;
   const hasTools = Array.isArray(tools) && tools.length > 0;
@@ -26068,15 +26071,19 @@ var LLMService = class {
     }
     const toolCallsAccumulator = /* @__PURE__ */ new Map();
     let reasoningSnapshot = "";
+    const reasoningEnabled = isReasoningEnabledForRequest(params, runtimeConfig);
     for await (const chunk of stream) {
       const choice = chunk.choices[0];
       if (!choice) continue;
       const delta = choice.delta;
       applyDeltaToolCallsToAccumulator(toolCallsAccumulator, delta?.tool_calls);
-      const rawReasoning = delta?.reasoning_content || delta?.reasoning || "";
-      const reasoning = extractStreamingDelta(reasoningSnapshot, rawReasoning);
-      if (reasoning) {
-        reasoningSnapshot += reasoning;
+      let reasoning = "";
+      if (reasoningEnabled) {
+        const rawReasoning = delta?.reasoning_content || delta?.reasoning || "";
+        reasoning = extractStreamingDelta(reasoningSnapshot, rawReasoning);
+        if (reasoning) {
+          reasoningSnapshot += reasoning;
+        }
       }
       const fullToolCalls = choice.finish_reason === "tool_calls" ? Array.from(toolCallsAccumulator.values()) : void 0;
       yield {
@@ -27802,20 +27809,23 @@ var BluMaTurnCoordinator = class {
     let toolCalls;
     let hasEmittedStart = false;
     let currentStreamPhase = "idle";
-    const stream = llmService.chatCompletionStream({
+    const llmUserContext = this.deps.getLlmUserContext();
+    const streamParams = {
       messages: contextWindow,
       temperature: 0,
       tools: this.deps.mcpClient.getAvailableTools(),
       parallel_tool_calls: true,
-      userContext: this.deps.getLlmUserContext()
-    });
+      userContext: llmUserContext
+    };
+    const reasoningEnabled = isReasoningEnabledForRequest(streamParams);
+    const stream = llmService.chatCompletionStream(streamParams);
     for await (const chunk of stream) {
       if (this.deps.isInterrupted()) {
         this.deps.eventBus.emit("stream_end", {});
         this.deps.eventBus.emit("backend_message", { type: "info", message: "Agent task cancelled by user." });
         return;
       }
-      if (chunk.reasoning) {
+      if (chunk.reasoning && reasoningEnabled) {
         if (!hasEmittedStart) {
           this.deps.eventBus.emit("stream_start", {});
           hasEmittedStart = true;
@@ -27936,7 +27946,11 @@ var BluMaTurnCoordinator = class {
     }
     let message2 = response.choices[0].message;
     message2 = ToolCallNormalizer.normalizeAssistantMessage(message2);
-    if (message2.reasoning_content || message2.reasoning) {
+    const reasoningEnabled = isReasoningEnabledForRequest({
+      messages: contextWindow,
+      userContext: this.deps.getLlmUserContext()
+    });
+    if (reasoningEnabled && (message2.reasoning_content || message2.reasoning)) {
       const reasoningText = message2.reasoning_content || message2.reasoning;
       this.deps.eventBus.emit("backend_message", {
         type: "reasoning",

package/package.json

@@ -1,6 +1,6 @@
 {
 	"name": "@nomad-e/bluma-cli",
-	"version": "0.26.1",
+	"version": "0.26.3",
 	"description": "BluMa independent agent for automation and advanced software engineering.",
 	"author": "Alex Fonseca",
 	"license": "Apache-2.0",