Updating ask sage models to support tool calling. Plan and Agent mode now work correctly.

core/llm/llms/Asksage.ts

@@ -4,6 +4,8 @@ import {
   CompletionOptions,
   LLMOptions,
   TextMessagePart,
+  Tool,
+  ToolCallDelta,
 } from "../../index.js";
 import { BaseLLM } from "../index.js";
 
@@ -61,15 +63,26 @@ interface AskSageRequestArgs {
   file?: unknown;
 }
 
+interface AskSageToolCall {
+  id: string;
+  type: "function";
+  function: {
+    name: string;
+    arguments: string;
+  };
+}
+
 interface AskSageResponse {
   text?: string;
   answer?: string;
   message?: string;
   status?: number | string;
   response?: unknown;
+  tool_calls?: AskSageToolCall[];
   choices?: Array<{
     message?: {
       content?: string;
+      tool_calls?: AskSageToolCall[];
     };
   }>;
 }
@@ -185,6 +198,17 @@ class Asksage extends BaseLLM {
     };
   }
 
+  protected _convertToolToAskSageTool(tool: Tool): AskSageTool {
+    return {
+      type: tool.type,
+      function: {
+        name: tool.function.name,
+        description: tool.function.description,
+        parameters: tool.function.parameters,
+      },
+    };
+  }
+
   protected _convertArgs(
     options: AskSageCompletionOptions,
     messages: ChatMessage[],
@@ -196,6 +220,14 @@ class Asksage extends BaseLLM {
       formattedMessage = messages.map(this._convertMessage);
     }
 
+    // Convert standard tools to AskSage format, or use askSageTools if provided
+    const tools =
+      options.tools?.map((tool) => this._convertToolToAskSageTool(tool)) ??
+      options.askSageTools;
+
+    // Map standard toolChoice to AskSage format, or use askSageToolChoice if provided
+    const toolChoice = options.toolChoice ?? options.askSageToolChoice;
+
     const args: AskSageRequestArgs = {
       message: formattedMessage,
       model: options.model,
@@ -207,10 +239,8 @@ class Asksage extends BaseLLM {
         options.systemPrompt ??
         process.env.ASKSAGE_SYSTEM_PROMPT ??
         "You are an expert software developer. You give helpful and concise responses.",
-      tools: options.askSageTools,
+      tools,
-      // enabled_mcp_tools: options.enabledMcpTools as string[] | undefined,
+      tool_choice: toolChoice,
-      // tools_to_execute: options.toolsToExecute as string[] | undefined,
-      tool_choice: options.askSageToolChoice,
       reasoning_effort: options.reasoningEffort as
         | "low"
         | "medium"
@@ -391,6 +421,22 @@ class Asksage extends BaseLLM {
 
       const data = (await response.json()) as AskSageResponse;
 
+      // Extract tool calls from response (check both top-level and choices format)
+      const rawToolCalls =
+        data.tool_calls || data.choices?.[0]?.message?.tool_calls;
+
+      // Convert to ToolCallDelta format if present
+      const toolCalls: ToolCallDelta[] | undefined = rawToolCalls?.map(
+        (tc) => ({
+          id: tc.id,
+          type: tc.type,
+          function: {
+            name: tc.function.name,
+            arguments: tc.function.arguments,
+          },
+        }),
+      );
+
       const assistantMessage: ChatMessage = {
         role: "assistant",
         content:
@@ -399,6 +445,7 @@ class Asksage extends BaseLLM {
           data.message ||
           data.choices?.[0]?.message?.content ||
           "",
+        ...(toolCalls && toolCalls.length > 0 ? { toolCalls } : {}),
       };
 
       yield assistantMessage;