Fix reasoning_content error in tool calls for kimi-k2.5 and other thinking models

.gitignore

@@ -12,6 +12,11 @@ build/
 /picoclaw-test
 cmd/**/workspace
 
+# Go debug binaries (created by VS Code Go extension)
+__debug_bin
+__debug_bin.exe
+__debug_bin*
+
 # Picoclaw specific
 
 # PicoClaw

pkg/providers/openai_compat/provider.go

@@ -286,6 +286,7 @@ func parseResponse(body []byte) (*LLMResponse, error) {
 type openaiMessage struct {
 	Role             string     `json:"role"`
 	Content          string     `json:"content"`
+	ReasoningContent string     `json:"reasoning_content,omitempty"`
 	ToolCalls        []ToolCall `json:"tool_calls,omitempty"`
 	ToolCallID       string     `json:"tool_call_id,omitempty"`
 }
@@ -299,6 +300,7 @@ func stripSystemParts(messages []Message) []openaiMessage {
 		out[i] = openaiMessage{
 			Role:             m.Role,
 			Content:          m.Content,
+			ReasoningContent: m.ReasoningContent,
 			ToolCalls:        m.ToolCalls,
 			ToolCallID:       m.ToolCallID,
 		}