diff --git a/.changeset/great-lamps-boil.md b/.changeset/great-lamps-boil.md
new file mode 100644
index 0000000..e4d6cd6
--- /dev/null
+++ b/.changeset/great-lamps-boil.md
@@ -0,0 +1,5 @@
+---
+"@mrck-labs/grid-core": patch
+---
+
+fix tool caling again
diff --git a/packages/core/src/factories/configurable-agent.factory.ts b/packages/core/src/factories/configurable-agent.factory.ts
index ae05179..922c995 100644
--- a/packages/core/src/factories/configurable-agent.factory.ts
+++ b/packages/core/src/factories/configurable-agent.factory.ts
@@ -774,8 +774,8 @@ export const createConfigurableAgent = async ({
         let fullText = "";
 
         // Call streaming LLM with tools
-        // Tool events are sent via sendUpdate callback in onStepFinish
-        const { textStream, generation } =
+        // Use fullStream to capture all events including tool calls
+        const { fullStream, generation } =
           await base.llmService.runStreamedLLMWithTools({
             messages: workingMessages,
             tools: availableTools,
@@ -795,21 +795,64 @@ export const createConfigurableAgent = async ({
             ...mergedLlmOptions,
           });
 
-        // Stream text chunks
-        for await (const chunk of textStream) {
-          fullText += chunk;
+        // Process the full stream which includes text, tool calls, and tool results
+        for await (const part of fullStream) {
+          // Handle text deltas
+          if (part.type === "text-delta") {
+            fullText += part.textDelta;
+
+            // Yield chunk to caller
+            yield {
+              type: "text_delta",
+              content: part.textDelta,
+            };
+
+            // Also send via sendUpdate for IPC
+            await sendUpdate({
+              type: "text_delta",
+              content: part.textDelta,
+            });
+          }
 
-          // Yield chunk to caller
-          yield {
-            type: "text_delta",
-            content: chunk,
-          };
+          // Handle tool calls - emit when tool is being called
+          if (part.type === "tool-call") {
+            const toolCallData = {
+              toolCallId: part.toolCallId,
+              toolName: part.toolName,
+              args: part.args,
+            };
+
+            yield {
+              type: "tool_execution",
+              content: JSON.stringify(toolCallData),
+              metadata: toolCallData,
+            };
+
+            await sendUpdate({
+              type: "tool_execution",
+              content: JSON.stringify(toolCallData),
+            });
+          }
 
-          // Also send via sendUpdate for IPC
-          await sendUpdate({
-            type: "text_delta",
-            content: chunk,
-          });
+          // Handle tool results - emit when tool returns
+          if (part.type === "tool-result") {
+            const toolResultData = {
+              toolCallId: part.toolCallId,
+              toolName: part.toolName,
+              result: part.result,
+            };
+
+            yield {
+              type: "tool_response",
+              content: JSON.stringify(toolResultData),
+              metadata: toolResultData,
+            };
+
+            await sendUpdate({
+              type: "tool_response",
+              content: JSON.stringify(toolResultData),
+            });
+          }
         }
 
         // Build final response
diff --git a/packages/core/src/services/base.llm.service.ts b/packages/core/src/services/base.llm.service.ts
index 218071a..54610e4 100644
--- a/packages/core/src/services/base.llm.service.ts
+++ b/packages/core/src/services/base.llm.service.ts
@@ -392,6 +392,7 @@ export const baseLLMService = (
     options: LLMServiceOptions & { tools?: any[] }
   ): Promise<{
     textStream: AsyncIterable<string>;
+    fullStream: AsyncIterable<any>;
     generation: any;
   }> => {
     const {
@@ -446,7 +447,7 @@ export const baseLLMService = (
       // Forward provider-specific options
       ...(providerOptions ? { providerOptions: providerOptions as any } : {}),
       onStepFinish: (step) => {
-        // Tool telemetry
+        // Tool telemetry - fires when each step completes
         step.content.forEach((content) => {
           if (content.type === "tool-call") {
             const sc: any = content;
@@ -463,35 +464,29 @@ export const baseLLMService = (
               toolName,
               args
             );
-            if (sendUpdate) {
-              sendUpdate({
-                type: "tool_execution",
-                content: JSON.stringify(content),
-              });
-            }
           }
 
           if (content.type === "tool-result") {
             const sc: any = content;
             const toolCallId = sc.toolCallId ?? sc.id ?? sc.callId;
-            const result = sc.result ?? sc.output ?? sc.data;
+            const resultData = sc.result ?? sc.output ?? sc.data;
             langfuse.endToolSpanForSession(
               options.context.sessionToken,
               toolCallId,
-              result
+              resultData
             );
-            if (sendUpdate) {
-              sendUpdate({
-                type: "tool_response",
-                content: JSON.stringify(content),
-              });
-            }
           }
         });
       },
     });
 
-    return { textStream: result.textStream, generation };
+    // Return both textStream and fullStream
+    // fullStream contains all events including tool-call and tool-result
+    return {
+      textStream: result.textStream,
+      fullStream: result.fullStream,
+      generation,
+    };
   };
 
   const isAvailable = async (): Promise<boolean> => {
diff --git a/packages/core/src/types/llm.types.ts b/packages/core/src/types/llm.types.ts
index d21e3ee..e7ad82f 100644
--- a/packages/core/src/types/llm.types.ts
+++ b/packages/core/src/types/llm.types.ts
@@ -123,6 +123,7 @@ export interface LLMService {
     options: LLMServiceOptions & { tools?: any[] }
   ): Promise<{
     textStream: AsyncIterable<string>;
+    fullStream: AsyncIterable<any>;
     generation: any;
   }>;