feat: add built-in agent(general-purpose)

2025-12-19 09:33:53 +00:00 · 2025-09-09 20:53:53 +08:00
parent 3c67dc0c0b
commit 549f296eb5
19 changed files with 896 additions and 228 deletions
--- a/docs/subagents.md
+++ b/docs/subagents.md
@@ -32,15 +32,19 @@ Subagents are independent AI assistants that:
 ### Quick Start

 1. **Create your first subagent**:
+
   ```
   /agents create
   ```
+
   Follow the guided wizard to create a specialized agent.

 2. **List existing agents**:
+
   ```
   /agents list
   ```
+
   View and manage your configured subagents.

 3. **Use subagents automatically**:
@@ -57,7 +61,6 @@ AI: I'll delegate this to your testing specialist subagent.
 [Returns with completed test files and execution summary]
 ```

-
 ## Management

 ### CLI Commands
@@ -69,6 +72,7 @@ Subagents are managed through the `/agents` slash command and its subcommands:
 Creates a new subagent through a guided step wizard.

 **Usage:**
+
 ```
 /agents create
 ```
@@ -78,6 +82,7 @@ Creates a new subagent through a guided step wizard.
 Opens an interactive management dialog for viewing and managing existing subagents.

 **Usage:**
+
 ```
 /agents list
 ```
@@ -141,15 +146,12 @@ Perfect for comprehensive test creation and test-driven development.
 name: testing-expert
 description: Writes comprehensive unit tests, integration tests, and handles test automation with best practices
 tools: read_file, write_file, read_many_files, run_shell_command
-modelConfig:
-  temp: 0.2
-runConfig:
-  max_time_minutes: 20
 ---

 You are a testing specialist focused on creating high-quality, maintainable tests.

 Your expertise includes:
+
 - Unit testing with appropriate mocking and isolation
 - Integration testing for component interactions
 - Test-driven development practices
@@ -157,6 +159,7 @@ Your expertise includes:
 - Performance and load testing when appropriate

 For each testing task:
+
 1. Analyze the code structure and dependencies
 2. Identify key functionality, edge cases, and error conditions
 3. Create comprehensive test suites with descriptive names
@@ -169,6 +172,7 @@ Focus on both positive and negative test cases.
 ```

 **Use Cases:**
+
 - "Write unit tests for the authentication service"
 - "Create integration tests for the payment processing workflow"
 - "Add test coverage for edge cases in the data validation module"
@@ -182,8 +186,6 @@ Specialized in creating clear, comprehensive documentation.
 name: documentation-writer
 description: Creates comprehensive documentation, README files, API docs, and user guides
 tools: read_file, write_file, read_many_files, web_search
-modelConfig:
-  temp: 0.4
 ---

 You are a technical documentation specialist for ${project_name}.
@@ -192,6 +194,7 @@ Your role is to create clear, comprehensive documentation that serves both
 developers and end users. Focus on:

 **For API Documentation:**
+
 - Clear endpoint descriptions with examples
 - Parameter details with types and constraints
 - Response format documentation
@@ -199,6 +202,7 @@ developers and end users. Focus on:
 - Authentication requirements

 **For User Documentation:**
+
 - Step-by-step instructions with screenshots when helpful
 - Installation and setup guides
 - Configuration options and examples
@@ -206,6 +210,7 @@ developers and end users. Focus on:
 - FAQ sections based on common user questions

 **For Developer Documentation:**
+
 - Architecture overviews and design decisions
 - Code examples that actually work
 - Contributing guidelines
@@ -216,6 +221,7 @@ the actual implementation. Use clear headings, bullet points, and examples.
 ```

 **Use Cases:**
+
 - "Create API documentation for the user management endpoints"
 - "Write a comprehensive README for this project"
 - "Document the deployment process with troubleshooting steps"
@@ -229,15 +235,12 @@ Focused on code quality, security, and best practices.
 name: code-reviewer
 description: Reviews code for best practices, security issues, performance, and maintainability
 tools: read_file, read_many_files
-modelConfig:
-  temp: 0.3
-runConfig:
-  max_time_minutes: 15
 ---

 You are an experienced code reviewer focused on quality, security, and maintainability.

 Review criteria:
+
 - **Code Structure**: Organization, modularity, and separation of concerns
 - **Performance**: Algorithmic efficiency and resource usage
 - **Security**: Vulnerability assessment and secure coding practices
@@ -247,6 +250,7 @@ Review criteria:
 - **Testing**: Test coverage and testability considerations

 Provide constructive feedback with:
+
 1. **Critical Issues**: Security vulnerabilities, major bugs
 2. **Important Improvements**: Performance issues, design problems
 3. **Minor Suggestions**: Style improvements, refactoring opportunities
@@ -257,6 +261,7 @@ Prioritize issues by impact and provide rationale for recommendations.
 ```

 **Use Cases:**
+
 - "Review this authentication implementation for security issues"
 - "Check the performance implications of this database query logic"
 - "Evaluate the code structure and suggest improvements"
@@ -272,13 +277,12 @@ Optimized for React development, hooks, and component patterns.
 name: react-specialist
 description: Expert in React development, hooks, component patterns, and modern React best practices
 tools: read_file, write_file, read_many_files, run_shell_command
-modelConfig:
-  temp: 0.3
 ---

 You are a React specialist with deep expertise in modern React development.

 Your expertise covers:
+
 - **Component Design**: Functional components, custom hooks, composition patterns
 - **State Management**: useState, useReducer, Context API, and external libraries
 - **Performance**: React.memo, useMemo, useCallback, code splitting
@@ -287,6 +291,7 @@ Your expertise covers:
 - **Modern Patterns**: Suspense, Error Boundaries, Concurrent Features

 For React tasks:
+
 1. Use functional components and hooks by default
 2. Implement proper TypeScript typing
 3. Follow React best practices and conventions
@@ -299,6 +304,7 @@ Focus on accessibility and user experience considerations.
 ```

 **Use Cases:**
+
 - "Create a reusable data table component with sorting and filtering"
 - "Implement a custom hook for API data fetching with caching"
 - "Refactor this class component to use modern React patterns"
@@ -312,13 +318,12 @@ Specialized in Python development, frameworks, and best practices.
 name: python-expert
 description: Expert in Python development, frameworks, testing, and Python-specific best practices
 tools: read_file, write_file, read_many_files, run_shell_command
-modelConfig:
-  temp: 0.3
 ---

 You are a Python expert with deep knowledge of the Python ecosystem.

 Your expertise includes:
+
 - **Core Python**: Pythonic patterns, data structures, algorithms
 - **Frameworks**: Django, Flask, FastAPI, SQLAlchemy
 - **Testing**: pytest, unittest, mocking, test-driven development
@@ -328,6 +333,7 @@ Your expertise includes:
 - **Code Quality**: PEP 8, type hints, linting with pylint/flake8

 For Python tasks:
+
 1. Follow PEP 8 style guidelines
 2. Use type hints for better code documentation
 3. Implement proper error handling with specific exceptions
@@ -340,6 +346,7 @@ Focus on writing clean, maintainable Python code that follows community standard
 ```

 **Use Cases:**
+
 - "Create a FastAPI service for user authentication with JWT tokens"
 - "Implement a data processing pipeline with pandas and error handling"
 - "Write a CLI tool using argparse with comprehensive help documentation"
@@ -353,6 +360,7 @@ Focus on writing clean, maintainable Python code that follows community standard
 Each subagent should have a clear, focused purpose.

 **✅ Good:**
+
 ```markdown
 ---
 name: testing-expert
@@ -361,6 +369,7 @@ description: Writes comprehensive unit tests and integration tests
 ```

 **❌ Avoid:**
+
 ```markdown
 ---
 name: general-helper
@@ -375,6 +384,7 @@ description: Helps with testing, documentation, code review, and deployment
 Define specific expertise areas rather than broad capabilities.

 **✅ Good:**
+
 ```markdown
 ---
 name: react-performance-optimizer
@@ -383,6 +393,7 @@ description: Optimizes React applications for performance using profiling and be
 ```

 **❌ Avoid:**
+
 ```markdown
 ---
 name: frontend-developer
@@ -397,11 +408,13 @@ description: Works on frontend development tasks
 Write descriptions that clearly indicate when to use the agent.

 **✅ Good:**
+
 ```markdown
 description: Reviews code for security vulnerabilities, performance issues, and maintainability concerns
 ```

 **❌ Avoid:**
+
 ```markdown
 description: A helpful code reviewer
 ```
@@ -413,8 +426,10 @@ description: A helpful code reviewer
 #### System Prompt Guidelines

 **Be Specific About Expertise:**
+
 ```markdown
 You are a Python testing specialist with expertise in:
+
 - pytest framework and fixtures
 - Mock objects and dependency injection
 - Test-driven development practices
@@ -422,8 +437,10 @@ You are a Python testing specialist with expertise in:
 ```

 **Include Step-by-Step Approaches:**
+
 ```markdown
 For each testing task:
+
 1. Analyze the code structure and dependencies
 2. Identify key functionality and edge cases
 3. Create comprehensive test suites with clear naming
@@ -432,8 +449,10 @@ For each testing task:
 ```

 **Specify Output Standards:**
+
 ```markdown
 Always follow these standards:
+
 - Use descriptive test names that explain the scenario
 - Include both positive and negative test cases
 - Add docstrings for complex test functions
--- a/packages/cli/src/ui/components/subagents/ActionSelectionStep.tsx
+++ b/packages/cli/src/ui/components/subagents/ActionSelectionStep.tsx
@@ -8,26 +8,37 @@ import { useState } from 'react';
 import { Box } from 'ink';
 import { RadioButtonSelect } from '../shared/RadioButtonSelect.js';
 import { MANAGEMENT_STEPS } from './types.js';
+import { SubagentConfig } from '@qwen-code/qwen-code-core';

 interface ActionSelectionStepProps {
+  selectedAgent: SubagentConfig | null;
  onNavigateToStep: (step: string) => void;
  onNavigateBack: () => void;
 }

 export const ActionSelectionStep = ({
+  selectedAgent,
  onNavigateToStep,
  onNavigateBack,
 }: ActionSelectionStepProps) => {
  const [selectedAction, setSelectedAction] = useState<
    'view' | 'edit' | 'delete' | null
  >(null);
-  const actions = [
+
+  // Filter actions based on whether the agent is built-in
+  const allActions = [
    { label: 'View Agent', value: 'view' as const },
    { label: 'Edit Agent', value: 'edit' as const },
    { label: 'Delete Agent', value: 'delete' as const },
    { label: 'Back', value: 'back' as const },
  ];

+  const actions = selectedAgent?.isBuiltin
+    ? allActions.filter(
+        (action) => action.value === 'view' || action.value === 'back',
+      )
+    : allActions;
+
  const handleActionSelect = (value: 'view' | 'edit' | 'delete' | 'back') => {
    if (value === 'back') {
      onNavigateBack();
--- a/packages/cli/src/ui/components/subagents/AgentSelectionStep.tsx
+++ b/packages/cli/src/ui/components/subagents/AgentSelectionStep.tsx
@@ -12,9 +12,10 @@ import { useKeypress } from '../../hooks/useKeypress.js';
 import { SubagentConfig } from '@qwen-code/qwen-code-core';

 interface NavigationState {
-  currentBlock: 'project' | 'user';
+  currentBlock: 'project' | 'user' | 'builtin';
  projectIndex: number;
  userIndex: number;
+  builtinIndex: number;
 }

 interface AgentSelectionStepProps {
@@ -30,6 +31,7 @@ export const AgentSelectionStep = ({
    currentBlock: 'project',
    projectIndex: 0,
    userIndex: 0,
+    builtinIndex: 0,
  });

  // Group agents by level
@@ -41,6 +43,10 @@ export const AgentSelectionStep = ({
    () => availableAgents.filter((agent) => agent.level === 'user'),
    [availableAgents],
  );
+  const builtinAgents = useMemo(
+    () => availableAgents.filter((agent) => agent.level === 'builtin'),
+    [availableAgents],
+  );
  const projectNames = useMemo(
    () => new Set(projectAgents.map((agent) => agent.name)),
    [projectAgents],
@@ -52,8 +58,10 @@ export const AgentSelectionStep = ({
      setNavigation((prev) => ({ ...prev, currentBlock: 'project' }));
    } else if (userAgents.length > 0) {
      setNavigation((prev) => ({ ...prev, currentBlock: 'user' }));
+    } else if (builtinAgents.length > 0) {
+      setNavigation((prev) => ({ ...prev, currentBlock: 'builtin' }));
    }
-  }, [projectAgents, userAgents]);
+  }, [projectAgents, userAgents, builtinAgents]);

  // Custom keyboard navigation
  useKeypress(
@@ -65,6 +73,13 @@ export const AgentSelectionStep = ({
          if (prev.currentBlock === 'project') {
            if (prev.projectIndex > 0) {
              return { ...prev, projectIndex: prev.projectIndex - 1 };
+            } else if (builtinAgents.length > 0) {
+              // Move to last item in builtin block
+              return {
+                ...prev,
+                currentBlock: 'builtin',
+                builtinIndex: builtinAgents.length - 1,
+              };
            } else if (userAgents.length > 0) {
              // Move to last item in user block
              return {
@@ -76,7 +91,7 @@ export const AgentSelectionStep = ({
              // Wrap to last item in project block
              return { ...prev, projectIndex: projectAgents.length - 1 };
            }
-          } else {
+          } else if (prev.currentBlock === 'user') {
            if (prev.userIndex > 0) {
              return { ...prev, userIndex: prev.userIndex - 1 };
            } else if (projectAgents.length > 0) {
@@ -86,10 +101,39 @@ export const AgentSelectionStep = ({
                currentBlock: 'project',
                projectIndex: projectAgents.length - 1,
              };
+            } else if (builtinAgents.length > 0) {
+              // Move to last item in builtin block
+              return {
+                ...prev,
+                currentBlock: 'builtin',
+                builtinIndex: builtinAgents.length - 1,
+              };
            } else {
              // Wrap to last item in user block
              return { ...prev, userIndex: userAgents.length - 1 };
            }
+          } else {
+            // builtin block
+            if (prev.builtinIndex > 0) {
+              return { ...prev, builtinIndex: prev.builtinIndex - 1 };
+            } else if (userAgents.length > 0) {
+              // Move to last item in user block
+              return {
+                ...prev,
+                currentBlock: 'user',
+                userIndex: userAgents.length - 1,
+              };
+            } else if (projectAgents.length > 0) {
+              // Move to last item in project block
+              return {
+                ...prev,
+                currentBlock: 'project',
+                projectIndex: projectAgents.length - 1,
+              };
+            } else {
+              // Wrap to last item in builtin block
+              return { ...prev, builtinIndex: builtinAgents.length - 1 };
+            }
          }
        });
      } else if (name === 'down' || name === 'j') {
@@ -100,13 +144,19 @@ export const AgentSelectionStep = ({
            } else if (userAgents.length > 0) {
              // Move to first item in user block
              return { ...prev, currentBlock: 'user', userIndex: 0 };
+            } else if (builtinAgents.length > 0) {
+              // Move to first item in builtin block
+              return { ...prev, currentBlock: 'builtin', builtinIndex: 0 };
            } else {
              // Wrap to first item in project block
              return { ...prev, projectIndex: 0 };
            }
-          } else {
+          } else if (prev.currentBlock === 'user') {
            if (prev.userIndex < userAgents.length - 1) {
              return { ...prev, userIndex: prev.userIndex + 1 };
+            } else if (builtinAgents.length > 0) {
+              // Move to first item in builtin block
+              return { ...prev, currentBlock: 'builtin', builtinIndex: 0 };
            } else if (projectAgents.length > 0) {
              // Move to first item in project block
              return { ...prev, currentBlock: 'project', projectIndex: 0 };
@@ -114,6 +164,20 @@ export const AgentSelectionStep = ({
              // Wrap to first item in user block
              return { ...prev, userIndex: 0 };
            }
+          } else {
+            // builtin block
+            if (prev.builtinIndex < builtinAgents.length - 1) {
+              return { ...prev, builtinIndex: prev.builtinIndex + 1 };
+            } else if (projectAgents.length > 0) {
+              // Move to first item in project block
+              return { ...prev, currentBlock: 'project', projectIndex: 0 };
+            } else if (userAgents.length > 0) {
+              // Move to first item in user block
+              return { ...prev, currentBlock: 'user', userIndex: 0 };
+            } else {
+              // Wrap to first item in builtin block
+              return { ...prev, builtinIndex: 0 };
+            }
          }
        });
      } else if (name === 'return' || name === 'space') {
@@ -121,9 +185,14 @@ export const AgentSelectionStep = ({
        let globalIndex: number;
        if (navigation.currentBlock === 'project') {
          globalIndex = navigation.projectIndex;
-        } else {
+        } else if (navigation.currentBlock === 'user') {
          // User agents come after project agents in the availableAgents array
          globalIndex = projectAgents.length + navigation.userIndex;
+        } else {
+          // builtin block
+          // Builtin agents come after project and user agents in the availableAgents array
+          globalIndex =
+            projectAgents.length + userAgents.length + navigation.builtinIndex;
        }

        if (globalIndex >= 0 && globalIndex < availableAgents.length) {
@@ -147,7 +216,11 @@ export const AgentSelectionStep = ({

  // Render custom radio button items
  const renderAgentItem = (
-    agent: { name: string; level: 'project' | 'user' },
+    agent: {
+      name: string;
+      level: 'project' | 'user' | 'builtin';
+      isBuiltin?: boolean;
+    },
    index: number,
    isSelected: boolean,
  ) => {
@@ -162,6 +235,12 @@ export const AgentSelectionStep = ({
        </Box>
        <Text color={textColor} wrap="truncate">
          {agent.name}
+          {agent.isBuiltin && (
+            <Text color={isSelected ? theme.text.accent : theme.text.secondary}>
+              {' '}
+              (built-in)
+            </Text>
+          )}
          {agent.level === 'user' && projectNames.has(agent.name) && (
            <Text color={isSelected ? theme.status.warning : Colors.Gray}>
              {' '}
@@ -176,7 +255,8 @@ export const AgentSelectionStep = ({
  // Calculate enabled agents count (excluding conflicted user-level agents)
  const enabledAgentsCount =
    projectAgents.length +
-    userAgents.filter((agent) => !projectNames.has(agent.name)).length;
+    userAgents.filter((agent) => !projectNames.has(agent.name)).length +
+    builtinAgents.length;

  return (
    <Box flexDirection="column">
@@ -199,7 +279,10 @@ export const AgentSelectionStep = ({

      {/* User Level Agents */}
      {userAgents.length > 0 && (
-        <Box flexDirection="column">
+        <Box
+          flexDirection="column"
+          marginBottom={builtinAgents.length > 0 ? 1 : 0}
+        >
          <Text color={theme.text.primary} bold>
            User Level ({userAgents[0].filePath.replace(/\/[^/]+$/, '')})
          </Text>
@@ -214,8 +297,27 @@ export const AgentSelectionStep = ({
        </Box>
      )}

+      {/* Built-in Agents */}
+      {builtinAgents.length > 0 && (
+        <Box flexDirection="column">
+          <Text color={theme.text.primary} bold>
+            Built-in Agents
+          </Text>
+          <Box marginTop={1} flexDirection="column">
+            {builtinAgents.map((agent, index) => {
+              const isSelected =
+                navigation.currentBlock === 'builtin' &&
+                navigation.builtinIndex === index;
+              return renderAgentItem(agent, index, isSelected);
+            })}
+          </Box>
+        </Box>
+      )}
+
      {/* Agent count summary */}
-      {(projectAgents.length > 0 || userAgents.length > 0) && (
+      {(projectAgents.length > 0 ||
+        userAgents.length > 0 ||
+        builtinAgents.length > 0) && (
        <Box marginTop={1}>
          <Text color={theme.text.secondary}>
            Using: {enabledAgentsCount} agents
--- a/packages/cli/src/ui/components/subagents/AgentViewerStep.tsx
+++ b/packages/cli/src/ui/components/subagents/AgentViewerStep.tsx
@@ -29,15 +29,6 @@ export const AgentViewerStep = ({ selectedAgent }: AgentViewerStepProps) => {
  return (
    <Box flexDirection="column" gap={1}>
      <Box flexDirection="column">
-        <Box>
-          <Text bold>Location: </Text>
-          <Text>
-            {agent.level === 'project'
-              ? 'Project Level (.qwen/agents/)'
-              : 'User Level (~/.qwen/agents/)'}
-          </Text>
-        </Box>
-
        <Box>
          <Text bold>File Path: </Text>
          <Text>{agent.filePath}</Text>
--- a/packages/cli/src/ui/components/subagents/AgentsManagerDialog.tsx
+++ b/packages/cli/src/ui/components/subagents/AgentsManagerDialog.tsx
@@ -50,14 +50,19 @@ export function AgentsManagerDialog({

    const manager = config.getSubagentManager();

-    // Load agents from both levels separately to show all agents including conflicts
-    const [projectAgents, userAgents] = await Promise.all([
+    // Load agents from all levels separately to show all agents including conflicts
+    const [projectAgents, userAgents, builtinAgents] = await Promise.all([
      manager.listSubagents({ level: 'project' }),
      manager.listSubagents({ level: 'user' }),
+      manager.listSubagents({ level: 'builtin' }),
    ]);

-    // Combine all agents (project and user level)
-    const allAgents = [...(projectAgents || []), ...(userAgents || [])];
+    // Combine all agents (project, user, and builtin level)
+    const allAgents = [
+      ...(projectAgents || []),
+      ...(userAgents || []),
+      ...(builtinAgents || []),
+    ];

    setAvailableAgents(allAgents);
  }, [config]);
@@ -208,7 +213,9 @@ export function AgentsManagerDialog({
          />
        );
      case MANAGEMENT_STEPS.ACTION_SELECTION:
-        return <ActionSelectionStep {...commonProps} />;
+        return (
+          <ActionSelectionStep selectedAgent={selectedAgent} {...commonProps} />
+        );
      case MANAGEMENT_STEPS.AGENT_VIEWER:
        return (
          <AgentViewerStep selectedAgent={selectedAgent} {...commonProps} />
--- a/packages/cli/src/ui/components/subagents/SubagentExecutionDisplay.tsx
+++ b/packages/cli/src/ui/components/subagents/SubagentExecutionDisplay.tsx
@@ -289,9 +289,7 @@ const ToolCallItem: React.FC<{
      <Box flexDirection="row">
        <Box minWidth={STATUS_INDICATOR_WIDTH}>{statusIcon}</Box>
        <Text wrap="truncate-end">
-          <Text>
-            {toolCall.name}
-          </Text>{' '}
+          <Text>{toolCall.name}</Text>{' '}
          <Text color={Colors.Gray}>{description}</Text>
          {toolCall.error && (
            <Text color={theme.status.error}> - {toolCall.error}</Text>
--- a/packages/cli/src/ui/components/subagents/ToolSelector.tsx
+++ b/packages/cli/src/ui/components/subagents/ToolSelector.tsx
@@ -65,7 +65,8 @@ export function ToolSelector({
        (tool) =>
          tool.kind === Kind.Read ||
          tool.kind === Kind.Search ||
-          tool.kind === Kind.Fetch,
+          tool.kind === Kind.Fetch ||
+          tool.kind === Kind.Think,
      )
      .map((tool) => tool.displayName)
      .sort();
@@ -75,8 +76,7 @@ export function ToolSelector({
        (tool) =>
          tool.kind === Kind.Edit ||
          tool.kind === Kind.Delete ||
-          tool.kind === Kind.Move ||
-          tool.kind === Kind.Think,
+          tool.kind === Kind.Move,
      )
      .map((tool) => tool.displayName)
      .sort();
--- a/packages/core/src/subagents/builtin-agents.test.ts
+++ b/packages/core/src/subagents/builtin-agents.test.ts
@@ -0,0 +1,95 @@
+/**
+ * @license
+ * Copyright 2025 Qwen
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import { describe, it, expect } from 'vitest';
+import { BuiltinAgentRegistry } from './builtin-agents.js';
+
+describe('BuiltinAgentRegistry', () => {
+  describe('getBuiltinAgents', () => {
+    it('should return array of builtin agents with correct properties', () => {
+      const agents = BuiltinAgentRegistry.getBuiltinAgents();
+
+      expect(agents).toBeInstanceOf(Array);
+      expect(agents.length).toBeGreaterThan(0);
+
+      agents.forEach((agent) => {
+        expect(agent).toMatchObject({
+          name: expect.any(String),
+          description: expect.any(String),
+          systemPrompt: expect.any(String),
+          level: 'builtin',
+          filePath: `<builtin:${agent.name}>`,
+          isBuiltin: true,
+        });
+      });
+    });
+
+    it('should include general-purpose agent', () => {
+      const agents = BuiltinAgentRegistry.getBuiltinAgents();
+      const generalAgent = agents.find(
+        (agent) => agent.name === 'general-purpose',
+      );
+
+      expect(generalAgent).toBeDefined();
+      expect(generalAgent?.description).toContain('General-purpose agent');
+    });
+  });
+
+  describe('getBuiltinAgent', () => {
+    it('should return correct agent for valid name', () => {
+      const agent = BuiltinAgentRegistry.getBuiltinAgent('general-purpose');
+
+      expect(agent).toMatchObject({
+        name: 'general-purpose',
+        level: 'builtin',
+        filePath: '<builtin:general-purpose>',
+        isBuiltin: true,
+      });
+    });
+
+    it('should return null for invalid name', () => {
+      expect(BuiltinAgentRegistry.getBuiltinAgent('invalid')).toBeNull();
+      expect(BuiltinAgentRegistry.getBuiltinAgent('')).toBeNull();
+    });
+  });
+
+  describe('isBuiltinAgent', () => {
+    it('should return true for valid builtin agent names', () => {
+      expect(BuiltinAgentRegistry.isBuiltinAgent('general-purpose')).toBe(true);
+    });
+
+    it('should return false for invalid names', () => {
+      expect(BuiltinAgentRegistry.isBuiltinAgent('invalid')).toBe(false);
+      expect(BuiltinAgentRegistry.isBuiltinAgent('')).toBe(false);
+    });
+  });
+
+  describe('getBuiltinAgentNames', () => {
+    it('should return array of agent names', () => {
+      const names = BuiltinAgentRegistry.getBuiltinAgentNames();
+
+      expect(names).toBeInstanceOf(Array);
+      expect(names).toContain('general-purpose');
+      expect(names.every((name) => typeof name === 'string')).toBe(true);
+    });
+  });
+
+  describe('consistency', () => {
+    it('should maintain consistency across all methods', () => {
+      const agents = BuiltinAgentRegistry.getBuiltinAgents();
+      const names = BuiltinAgentRegistry.getBuiltinAgentNames();
+
+      // Names should match agents
+      expect(names).toEqual(agents.map((agent) => agent.name));
+
+      // Each name should be valid
+      names.forEach((name) => {
+        expect(BuiltinAgentRegistry.isBuiltinAgent(name)).toBe(true);
+        expect(BuiltinAgentRegistry.getBuiltinAgent(name)).toBeDefined();
+      });
+    });
+  });
+});
--- a/packages/core/src/subagents/builtin-agents.ts
+++ b/packages/core/src/subagents/builtin-agents.ts
@@ -0,0 +1,95 @@
+/**
+ * @license
+ * Copyright 2025 Qwen
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import { SubagentConfig } from './types.js';
+
+/**
+ * Registry of built-in subagents that are always available to all users.
+ * These agents are embedded in the codebase and cannot be modified or deleted.
+ */
+export class BuiltinAgentRegistry {
+  private static readonly BUILTIN_AGENTS: Array<
+    Omit<SubagentConfig, 'level' | 'filePath'>
+  > = [
+    {
+      name: 'general-purpose',
+      description:
+        'General-purpose agent for researching complex questions, searching for code, and executing multi-step tasks. When you are searching for a keyword or file and are not confident that you will find the right match in the first few tries use this agent to perform the search for you.',
+      systemPrompt: `You are a general-purpose research and code analysis agent. Given the user's message, you should use the tools available to complete the task. Do what has been asked; nothing more, nothing less. When you complete the task simply respond with a detailed writeup.
+
+Your strengths:
+- Searching for code, configurations, and patterns across large codebases
+- Analyzing multiple files to understand system architecture
+- Investigating complex questions that require exploring many files
+- Performing multi-step research tasks
+
+Guidelines:
+- For file searches: Use Grep or Glob when you need to search broadly. Use Read when you know the specific file path.
+- For analysis: Start broad and narrow down. Use multiple search strategies if the first doesn't yield results.
+- Be thorough: Check multiple locations, consider different naming conventions, look for related files.
+- NEVER create files unless they're absolutely necessary for achieving your goal. ALWAYS prefer editing an existing file to creating a new one.
+- NEVER proactively create documentation files (*.md) or README files. Only create documentation files if explicitly requested.
+- In your final response always share relevant file names and code snippets. Any file paths you return in your response MUST be absolute. Do NOT use relative paths.
+- For clear communication, avoid using emojis.
+
+
+Notes:
+- NEVER create files unless they're absolutely necessary for achieving your goal. ALWAYS prefer editing an existing file to creating a new one.
+- NEVER proactively create documentation files (*.md) or README files. Only create documentation files if explicitly requested by the User.
+- In your final response always share relevant file names and code snippets. Any file paths you return in your response MUST be absolute. Do NOT use relative paths.
+- For clear communication with the user the assistant MUST avoid using emojis.`,
+    },
+  ];
+
+  /**
+   * Gets all built-in agent configurations.
+   * @returns Array of built-in subagent configurations
+   */
+  static getBuiltinAgents(): SubagentConfig[] {
+    return this.BUILTIN_AGENTS.map((agent) => ({
+      ...agent,
+      level: 'builtin' as const,
+      filePath: `<builtin:${agent.name}>`,
+      isBuiltin: true,
+    }));
+  }
+
+  /**
+   * Gets a specific built-in agent by name.
+   * @param name - Name of the built-in agent
+   * @returns Built-in agent configuration or null if not found
+   */
+  static getBuiltinAgent(name: string): SubagentConfig | null {
+    const agent = this.BUILTIN_AGENTS.find((a) => a.name === name);
+    if (!agent) {
+      return null;
+    }
+
+    return {
+      ...agent,
+      level: 'builtin' as const,
+      filePath: `<builtin:${name}>`,
+      isBuiltin: true,
+    };
+  }
+
+  /**
+   * Checks if an agent name corresponds to a built-in agent.
+   * @param name - Agent name to check
+   * @returns True if the name is a built-in agent
+   */
+  static isBuiltinAgent(name: string): boolean {
+    return this.BUILTIN_AGENTS.some((agent) => agent.name === name);
+  }
+
+  /**
+   * Gets the names of all built-in agents.
+   * @returns Array of built-in agent names
+   */
+  static getBuiltinAgentNames(): string[] {
+    return this.BUILTIN_AGENTS.map((agent) => agent.name);
+  }
+}
--- a/packages/core/src/subagents/index.ts
+++ b/packages/core/src/subagents/index.ts
@@ -33,6 +33,9 @@ export type {

 export { SubagentError } from './types.js';

+// Built-in agents registry
+export { BuiltinAgentRegistry } from './builtin-agents.js';
+
 // Validation system
 export { SubagentValidator } from './validation.js';

@@ -70,4 +73,3 @@ export type {
  SubagentStatsSummary,
  ToolUsageStats,
 } from './subagent-statistics.js';
-export { formatCompact, formatDetailed } from './subagent-result-format.js';
--- a/packages/core/src/subagents/subagent-manager.test.ts
+++ b/packages/core/src/subagents/subagent-manager.test.ts
@@ -584,11 +584,12 @@ System prompt 3`);
    it('should list subagents from both levels', async () => {
      const subagents = await manager.listSubagents();

-      expect(subagents).toHaveLength(3); // agent1 (project takes precedence), agent2, agent3
+      expect(subagents).toHaveLength(4); // agent1 (project takes precedence), agent2, agent3, general-purpose (built-in)
      expect(subagents.map((s) => s.name)).toEqual([
        'agent1',
        'agent2',
        'agent3',
+        'general-purpose',
      ]);
    });

@@ -615,7 +616,7 @@ System prompt 3`);
      });

      const names = subagents.map((s) => s.name);
-      expect(names).toEqual(['agent1', 'agent2', 'agent3']);
+      expect(names).toEqual(['agent1', 'agent2', 'agent3', 'general-purpose']);
    });

    it('should handle empty directories', async () => {
@@ -626,7 +627,9 @@ System prompt 3`);

      const subagents = await manager.listSubagents();

-      expect(subagents).toHaveLength(0);
+      expect(subagents).toHaveLength(1); // Only built-in agents remain
+      expect(subagents[0].name).toBe('general-purpose');
+      expect(subagents[0].level).toBe('builtin');
    });

    it('should handle directory read errors', async () => {
@@ -636,7 +639,9 @@ System prompt 3`);

      const subagents = await manager.listSubagents();

-      expect(subagents).toHaveLength(0);
+      expect(subagents).toHaveLength(1); // Only built-in agents remain
+      expect(subagents[0].name).toBe('general-purpose');
+      expect(subagents[0].level).toBe('builtin');
    });

    it('should skip invalid subagent files', async () => {
@@ -656,7 +661,7 @@ System prompt 3`);

      const subagents = await manager.listSubagents();

-      expect(subagents).toHaveLength(1);
+      expect(subagents).toHaveLength(2); // 1 valid file + 1 built-in agent
      expect(consoleSpy).toHaveBeenCalledWith(
        expect.stringContaining('Skipping invalid subagent file'),
      );
--- a/packages/core/src/subagents/subagent-manager.ts
+++ b/packages/core/src/subagents/subagent-manager.ts
@@ -29,6 +29,7 @@ import {
 import { SubagentValidator } from './validation.js';
 import { SubAgentScope } from './subagent.js';
 import { Config } from '../config/config.js';
+import { BuiltinAgentRegistry } from './builtin-agents.js';

 const QWEN_CONFIG_DIR = '.qwen';
 const AGENT_CONFIG_DIR = 'agents';
@@ -104,7 +105,7 @@ export class SubagentManager {
  /**
   * Loads a subagent configuration by name.
   * If level is specified, only searches that level.
-   * If level is omitted, searches project-level first, then user-level.
+   * If level is omitted, searches project-level first, then user-level, then built-in.
   *
   * @param name - Name of the subagent to load
   * @param level - Optional level to limit search to specific level
@@ -116,6 +117,10 @@ export class SubagentManager {
  ): Promise<SubagentConfig | null> {
    if (level) {
      // Search only the specified level
+      if (level === 'builtin') {
+        return BuiltinAgentRegistry.getBuiltinAgent(name);
+      }
+
      const path = this.getSubagentPath(name, level);
      try {
        const config = await this.parseSubagentFile(path);
@@ -140,9 +145,11 @@ export class SubagentManager {
      const config = await this.parseSubagentFile(userPath);
      return config;
    } catch (_error) {
-      // Not found at either level
-      return null;
+      // Continue to built-in agents
    }
+
+    // Try built-in agents as fallback
+    return BuiltinAgentRegistry.getBuiltinAgent(name);
  }

  /**
@@ -166,6 +173,15 @@ export class SubagentManager {
      );
    }

+    // Prevent updating built-in agents
+    if (existing.isBuiltin) {
+      throw new SubagentError(
+        `Cannot update built-in subagent "${name}"`,
+        SubagentErrorCode.INVALID_CONFIG,
+        name,
+      );
+    }
+
    // Merge updates with existing configuration
    const updatedConfig = this.mergeConfigurations(existing, updates);

@@ -194,12 +210,26 @@ export class SubagentManager {
   * @throws SubagentError if deletion fails
   */
  async deleteSubagent(name: string, level?: SubagentLevel): Promise<void> {
+    // Check if it's a built-in agent first
+    if (BuiltinAgentRegistry.isBuiltinAgent(name)) {
+      throw new SubagentError(
+        `Cannot delete built-in subagent "${name}"`,
+        SubagentErrorCode.INVALID_CONFIG,
+        name,
+      );
+    }
+
    const levelsToCheck: SubagentLevel[] = level
      ? [level]
      : ['project', 'user'];
    let deleted = false;

    for (const currentLevel of levelsToCheck) {
+      // Skip builtin level for deletion
+      if (currentLevel === 'builtin') {
+        continue;
+      }
+
      const filePath = this.getSubagentPath(name, currentLevel);

      try {
@@ -233,14 +263,14 @@ export class SubagentManager {

    const levelsToCheck: SubagentLevel[] = options.level
      ? [options.level]
-      : ['project', 'user'];
+      : ['project', 'user', 'builtin'];

-    // Collect subagents from each level (project takes precedence)
+    // Collect subagents from each level (project takes precedence over user, user takes precedence over builtin)
    for (const level of levelsToCheck) {
      const levelSubagents = await this.listSubagentsAtLevel(level);

      for (const subagent of levelSubagents) {
-        // Skip if we've already seen this name (project takes precedence)
+        // Skip if we've already seen this name (precedence: project > user > builtin)
        if (seenNames.has(subagent.name)) {
          continue;
        }
@@ -267,11 +297,12 @@ export class SubagentManager {
          case 'name':
            comparison = a.name.localeCompare(b.name);
            break;
-          case 'level':
-            // Project comes before user
-            comparison =
-              a.level === 'project' ? -1 : b.level === 'project' ? 1 : 0;
+          case 'level': {
+            // Project comes before user, user comes before builtin
+            const levelOrder = { project: 0, user: 1, builtin: 2 };
+            comparison = levelOrder[a.level] - levelOrder[b.level];
            break;
+          }
          default:
            comparison = 0;
            break;
@@ -605,6 +636,10 @@ export class SubagentManager {
   * @returns Absolute file path
   */
  getSubagentPath(name: string, level: SubagentLevel): string {
+    if (level === 'builtin') {
+      return `<builtin:${name}>`;
+    }
+
    const baseDir =
      level === 'project'
        ? path.join(
@@ -626,6 +661,11 @@ export class SubagentManager {
  private async listSubagentsAtLevel(
    level: SubagentLevel,
  ): Promise<SubagentConfig[]> {
+    // Handle built-in agents
+    if (level === 'builtin') {
+      return BuiltinAgentRegistry.getBuiltinAgents();
+    }
+
    const baseDir =
      level === 'project'
        ? path.join(
--- a/packages/core/src/subagents/subagent-result-format.ts
+++ b/packages/core/src/subagents/subagent-result-format.ts
@@ -1,156 +0,0 @@
-/**
- * @license
- * Copyright 2025 Qwen
- * SPDX-License-Identifier: Apache-2.0
- */
-
-import { SubagentStatsSummary } from './subagent-statistics.js';
-
-function fmtDuration(ms: number): string {
-  if (ms < 1000) return `${Math.round(ms)}ms`;
-  if (ms < 60000) return `${(ms / 1000).toFixed(1)}s`;
-  if (ms < 3600000) {
-    const m = Math.floor(ms / 60000);
-    const s = Math.floor((ms % 60000) / 1000);
-    return `${m}m ${s}s`;
-  }
-  const h = Math.floor(ms / 3600000);
-  const m = Math.floor((ms % 3600000) / 60000);
-  return `${h}h ${m}m`;
-}
-
-export function formatCompact(
-  stats: SubagentStatsSummary,
-  taskDesc: string,
-): string {
-  const sr =
-    stats.totalToolCalls > 0
-      ? (stats.successRate ??
-        (stats.successfulToolCalls / stats.totalToolCalls) * 100)
-      : 0;
-  const lines = [
-    `📋 Task Completed: ${taskDesc}`,
-    `🔧 Tool Usage: ${stats.totalToolCalls} calls${stats.totalToolCalls ? `, ${sr.toFixed(1)}% success` : ''}`,
-    `⏱️ Duration: ${fmtDuration(stats.totalDurationMs)} | 🔁 Rounds: ${stats.rounds}`,
-  ];
-  if (typeof stats.totalTokens === 'number') {
-    lines.push(
-      `🔢 Tokens: ${stats.totalTokens.toLocaleString()}${stats.inputTokens || stats.outputTokens ? ` (in ${stats.inputTokens ?? 0}, out ${stats.outputTokens ?? 0})` : ''}`,
-    );
-  }
-  return lines.join('\n');
-}
-
-export function formatDetailed(
-  stats: SubagentStatsSummary,
-  taskDesc: string,
-): string {
-  const sr =
-    stats.totalToolCalls > 0
-      ? (stats.successRate ??
-        (stats.successfulToolCalls / stats.totalToolCalls) * 100)
-      : 0;
-  const lines: string[] = [];
-  lines.push(`📋 Task Completed: ${taskDesc}`);
-  lines.push(
-    `⏱️ Duration: ${fmtDuration(stats.totalDurationMs)} | 🔁 Rounds: ${stats.rounds}`,
-  );
-  // Quality indicator
-  let quality = 'Poor execution';
-  if (sr >= 95) quality = 'Excellent execution';
-  else if (sr >= 85) quality = 'Good execution';
-  else if (sr >= 70) quality = 'Fair execution';
-  lines.push(`✅ Quality: ${quality} (${sr.toFixed(1)}% tool success)`);
-  // Speed category
-  const d = stats.totalDurationMs;
-  let speed = 'Long execution - consider breaking down tasks';
-  if (d < 10_000) speed = 'Fast completion - under 10 seconds';
-  else if (d < 60_000) speed = 'Good speed - under a minute';
-  else if (d < 300_000) speed = 'Moderate duration - a few minutes';
-  lines.push(`🚀 Speed: ${speed}`);
-  lines.push(
-    `🔧 Tools: ${stats.totalToolCalls} calls, ${sr.toFixed(1)}% success (${stats.successfulToolCalls} ok, ${stats.failedToolCalls} failed)`,
-  );
-  if (typeof stats.totalTokens === 'number') {
-    lines.push(
-      `🔢 Tokens: ${stats.totalTokens.toLocaleString()} (in ${stats.inputTokens ?? 0}, out ${stats.outputTokens ?? 0})`,
-    );
-  }
-  if (stats.toolUsage && stats.toolUsage.length) {
-    const sorted = [...stats.toolUsage]
-      .sort((a, b) => b.count - a.count)
-      .slice(0, 5);
-    lines.push('\nTop tools:');
-    for (const t of sorted) {
-      const avg =
-        typeof t.averageDurationMs === 'number'
-          ? `, avg ${fmtDuration(Math.round(t.averageDurationMs))}`
-          : '';
-      lines.push(
-        ` - ${t.name}: ${t.count} calls (${t.success} ok, ${t.failure} fail${avg}${t.lastError ? `, last error: ${t.lastError}` : ''})`,
-      );
-    }
-  }
-  const tips = generatePerformanceTips(stats);
-  if (tips.length) {
-    lines.push('\n💡 Performance Insights:');
-    for (const tip of tips.slice(0, 3)) lines.push(` - ${tip}`);
-  }
-  return lines.join('\n');
-}
-
-export function generatePerformanceTips(stats: SubagentStatsSummary): string[] {
-  const tips: string[] = [];
-  const totalCalls = stats.totalToolCalls;
-  const sr =
-    stats.totalToolCalls > 0
-      ? (stats.successRate ??
-        (stats.successfulToolCalls / stats.totalToolCalls) * 100)
-      : 0;
-
-  // High failure rate
-  if (sr < 80)
-    tips.push('Low tool success rate - review inputs and error messages');
-
-  // Long duration
-  if (stats.totalDurationMs > 60_000)
-    tips.push('Long execution time - consider breaking down complex tasks');
-
-  // Token usage
-  if (typeof stats.totalTokens === 'number' && stats.totalTokens > 100_000) {
-    tips.push(
-      'High token usage - consider optimizing prompts or narrowing scope',
-    );
-  }
-  if (typeof stats.totalTokens === 'number' && totalCalls > 0) {
-    const avgTokPerCall = stats.totalTokens / totalCalls;
-    if (avgTokPerCall > 5_000)
-      tips.push(
-        `High token usage per tool call (~${Math.round(avgTokPerCall)} tokens/call)`,
-      );
-  }
-
-  // Network failures
-  const isNetworkTool = (name: string) => /web|fetch|search/i.test(name);
-  const hadNetworkFailure = (stats.toolUsage || []).some(
-    (t) =>
-      isNetworkTool(t.name) &&
-      t.lastError &&
-      /timeout|network/i.test(t.lastError),
-  );
-  if (hadNetworkFailure)
-    tips.push(
-      'Network operations had failures - consider increasing timeout or checking connectivity',
-    );
-
-  // Slow tools
-  const slow = (stats.toolUsage || [])
-    .filter((t) => (t.averageDurationMs ?? 0) > 10_000)
-    .sort((a, b) => (b.averageDurationMs ?? 0) - (a.averageDurationMs ?? 0));
-  if (slow.length)
-    tips.push(
-      `Consider optimizing ${slow[0].name} operations (avg ${fmtDuration(Math.round(slow[0].averageDurationMs!))})`,
-    );
-
-  return tips;
-}
--- a/packages/core/src/subagents/subagent-statistics.test.ts
+++ b/packages/core/src/subagents/subagent-statistics.test.ts
@@ -0,0 +1,309 @@
+/**
+ * @license
+ * Copyright 2025 Qwen
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import { describe, it, expect, beforeEach } from 'vitest';
+import { SubagentStatistics } from './subagent-statistics.js';
+
+describe('SubagentStatistics', () => {
+  let stats: SubagentStatistics;
+  const baseTime = 1000000000000; // Fixed timestamp for consistent testing
+
+  beforeEach(() => {
+    stats = new SubagentStatistics();
+  });
+
+  describe('basic statistics tracking', () => {
+    it('should track execution time', () => {
+      stats.start(baseTime);
+      const summary = stats.getSummary(baseTime + 5000);
+
+      expect(summary.totalDurationMs).toBe(5000);
+    });
+
+    it('should track rounds', () => {
+      stats.setRounds(3);
+      const summary = stats.getSummary();
+
+      expect(summary.rounds).toBe(3);
+    });
+
+    it('should track tool calls', () => {
+      stats.recordToolCall('file_read', true, 100);
+      stats.recordToolCall('web_search', false, 200, 'Network timeout');
+
+      const summary = stats.getSummary();
+      expect(summary.totalToolCalls).toBe(2);
+      expect(summary.successfulToolCalls).toBe(1);
+      expect(summary.failedToolCalls).toBe(1);
+      expect(summary.successRate).toBe(50);
+    });
+
+    it('should track tokens', () => {
+      stats.recordTokens(1000, 500);
+      stats.recordTokens(200, 100);
+
+      const summary = stats.getSummary();
+      expect(summary.inputTokens).toBe(1200);
+      expect(summary.outputTokens).toBe(600);
+      expect(summary.totalTokens).toBe(1800);
+    });
+  });
+
+  describe('tool usage statistics', () => {
+    it('should track individual tool usage', () => {
+      stats.recordToolCall('file_read', true, 100);
+      stats.recordToolCall('file_read', false, 150, 'Permission denied');
+      stats.recordToolCall('web_search', true, 300);
+
+      const summary = stats.getSummary();
+      const fileReadTool = summary.toolUsage.find(
+        (t) => t.name === 'file_read',
+      );
+      const webSearchTool = summary.toolUsage.find(
+        (t) => t.name === 'web_search',
+      );
+
+      expect(fileReadTool).toEqual({
+        name: 'file_read',
+        count: 2,
+        success: 1,
+        failure: 1,
+        lastError: 'Permission denied',
+        totalDurationMs: 250,
+        averageDurationMs: 125,
+      });
+
+      expect(webSearchTool).toEqual({
+        name: 'web_search',
+        count: 1,
+        success: 1,
+        failure: 0,
+        lastError: undefined,
+        totalDurationMs: 300,
+        averageDurationMs: 300,
+      });
+    });
+  });
+
+  describe('formatCompact', () => {
+    it('should format basic execution summary', () => {
+      stats.start(baseTime);
+      stats.setRounds(2);
+      stats.recordToolCall('file_read', true, 100);
+      stats.recordTokens(1000, 500);
+
+      const result = stats.formatCompact('Test task', baseTime + 5000);
+
+      expect(result).toContain('📋 Task Completed: Test task');
+      expect(result).toContain('🔧 Tool Usage: 1 calls, 100.0% success');
+      expect(result).toContain('⏱️ Duration: 5.0s | 🔁 Rounds: 2');
+      expect(result).toContain('🔢 Tokens: 1,500 (in 1000, out 500)');
+    });
+
+    it('should handle zero tool calls', () => {
+      stats.start(baseTime);
+
+      const result = stats.formatCompact('Empty task', baseTime + 1000);
+
+      expect(result).toContain('🔧 Tool Usage: 0 calls');
+      expect(result).not.toContain('% success');
+    });
+
+    it('should show zero tokens when no tokens recorded', () => {
+      stats.start(baseTime);
+      stats.recordToolCall('test', true, 100);
+
+      const result = stats.formatCompact('No tokens task', baseTime + 1000);
+
+      expect(result).toContain('🔢 Tokens: 0');
+    });
+  });
+
+  describe('formatDetailed', () => {
+    beforeEach(() => {
+      stats.start(baseTime);
+      stats.setRounds(3);
+      stats.recordToolCall('file_read', true, 100);
+      stats.recordToolCall('file_read', true, 150);
+      stats.recordToolCall('web_search', false, 2000, 'Network timeout');
+      stats.recordTokens(2000, 1000);
+    });
+
+    it('should include quality assessment', () => {
+      const result = stats.formatDetailed('Complex task', baseTime + 30000);
+
+      expect(result).toContain(
+        '✅ Quality: Poor execution (66.7% tool success)',
+      );
+    });
+
+    it('should include speed assessment', () => {
+      const result = stats.formatDetailed('Fast task', baseTime + 5000);
+
+      expect(result).toContain('🚀 Speed: Fast completion - under 10 seconds');
+    });
+
+    it('should show top tools', () => {
+      const result = stats.formatDetailed('Tool-heavy task', baseTime + 15000);
+
+      expect(result).toContain('Top tools:');
+      expect(result).toContain('- file_read: 2 calls (2 ok, 0 fail');
+      expect(result).toContain('- web_search: 1 calls (0 ok, 1 fail');
+      expect(result).toContain('last error: Network timeout');
+    });
+
+    it('should include performance insights', () => {
+      const result = stats.formatDetailed('Slow task', baseTime + 120000);
+
+      expect(result).toContain('💡 Performance Insights:');
+      expect(result).toContain(
+        'Long execution time - consider breaking down complex tasks',
+      );
+    });
+  });
+
+  describe('quality categories', () => {
+    it('should categorize excellent execution', () => {
+      stats.recordToolCall('test', true, 100);
+      stats.recordToolCall('test', true, 100);
+
+      const result = stats.formatDetailed('Perfect task');
+      expect(result).toContain('Excellent execution (100.0% tool success)');
+    });
+
+    it('should categorize good execution', () => {
+      // Need 85% success rate for "Good execution" - 17 success, 3 failures = 85%
+      for (let i = 0; i < 17; i++) {
+        stats.recordToolCall('test', true, 100);
+      }
+      for (let i = 0; i < 3; i++) {
+        stats.recordToolCall('test', false, 100);
+      }
+
+      const result = stats.formatDetailed('Good task');
+      expect(result).toContain('Good execution (85.0% tool success)');
+    });
+
+    it('should categorize poor execution', () => {
+      stats.recordToolCall('test', false, 100);
+      stats.recordToolCall('test', false, 100);
+
+      const result = stats.formatDetailed('Poor task');
+      expect(result).toContain('Poor execution (0.0% tool success)');
+    });
+  });
+
+  describe('speed categories', () => {
+    it('should categorize fast completion', () => {
+      stats.start(baseTime);
+      const result = stats.formatDetailed('Fast task', baseTime + 5000);
+      expect(result).toContain('Fast completion - under 10 seconds');
+    });
+
+    it('should categorize good speed', () => {
+      stats.start(baseTime);
+      const result = stats.formatDetailed('Medium task', baseTime + 30000);
+      expect(result).toContain('Good speed - under a minute');
+    });
+
+    it('should categorize moderate duration', () => {
+      stats.start(baseTime);
+      const result = stats.formatDetailed('Slow task', baseTime + 120000);
+      expect(result).toContain('Moderate duration - a few minutes');
+    });
+
+    it('should categorize long execution', () => {
+      stats.start(baseTime);
+      const result = stats.formatDetailed('Very slow task', baseTime + 600000);
+      expect(result).toContain('Long execution - consider breaking down tasks');
+    });
+  });
+
+  describe('performance tips', () => {
+    it('should suggest reviewing low success rate', () => {
+      stats.recordToolCall('test', false, 100);
+      stats.recordToolCall('test', false, 100);
+      stats.recordToolCall('test', true, 100);
+
+      const result = stats.formatDetailed('Failing task');
+      expect(result).toContain(
+        'Low tool success rate - review inputs and error messages',
+      );
+    });
+
+    it('should suggest breaking down long tasks', () => {
+      stats.start(baseTime);
+
+      const result = stats.formatDetailed('Long task', baseTime + 120000);
+      expect(result).toContain(
+        'Long execution time - consider breaking down complex tasks',
+      );
+    });
+
+    it('should suggest optimizing high token usage', () => {
+      stats.recordTokens(80000, 30000);
+
+      const result = stats.formatDetailed('Token-heavy task');
+      expect(result).toContain(
+        'High token usage - consider optimizing prompts or narrowing scope',
+      );
+    });
+
+    it('should identify high token usage per call', () => {
+      stats.recordToolCall('test', true, 100);
+      stats.recordTokens(6000, 0);
+
+      const result = stats.formatDetailed('Verbose task');
+      expect(result).toContain(
+        'High token usage per tool call (~6000 tokens/call)',
+      );
+    });
+
+    it('should identify network failures', () => {
+      stats.recordToolCall('web_search', false, 100, 'Network timeout');
+
+      const result = stats.formatDetailed('Network task');
+      expect(result).toContain(
+        'Network operations had failures - consider increasing timeout or checking connectivity',
+      );
+    });
+
+    it('should identify slow tools', () => {
+      stats.recordToolCall('slow_tool', true, 15000);
+
+      const result = stats.formatDetailed('Slow tool task');
+      expect(result).toContain(
+        'Consider optimizing slow_tool operations (avg 15.0s)',
+      );
+    });
+  });
+
+  describe('duration formatting', () => {
+    it('should format milliseconds', () => {
+      stats.start(baseTime);
+      const result = stats.formatCompact('Quick task', baseTime + 500);
+      expect(result).toContain('500ms');
+    });
+
+    it('should format seconds', () => {
+      stats.start(baseTime);
+      const result = stats.formatCompact('Second task', baseTime + 2500);
+      expect(result).toContain('2.5s');
+    });
+
+    it('should format minutes and seconds', () => {
+      stats.start(baseTime);
+      const result = stats.formatCompact('Minute task', baseTime + 125000);
+      expect(result).toContain('2m 5s');
+    });
+
+    it('should format hours and minutes', () => {
+      stats.start(baseTime);
+      const result = stats.formatCompact('Hour task', baseTime + 4500000);
+      expect(result).toContain('1h 15m');
+    });
+  });
+});
--- a/packages/core/src/subagents/subagent-statistics.ts
+++ b/packages/core/src/subagents/subagent-statistics.ts
@@ -102,4 +102,149 @@ export class SubagentStatistics {
      toolUsage: Array.from(this.toolUsage.values()),
    };
  }
+
+  formatCompact(taskDesc: string, now = Date.now()): string {
+    const stats = this.getSummary(now);
+    const sr =
+      stats.totalToolCalls > 0
+        ? (stats.successRate ??
+          (stats.successfulToolCalls / stats.totalToolCalls) * 100)
+        : 0;
+    const lines = [
+      `📋 Task Completed: ${taskDesc}`,
+      `🔧 Tool Usage: ${stats.totalToolCalls} calls${stats.totalToolCalls ? `, ${sr.toFixed(1)}% success` : ''}`,
+      `⏱️ Duration: ${this.fmtDuration(stats.totalDurationMs)} | 🔁 Rounds: ${stats.rounds}`,
+    ];
+    if (typeof stats.totalTokens === 'number') {
+      lines.push(
+        `🔢 Tokens: ${stats.totalTokens.toLocaleString()}${stats.inputTokens || stats.outputTokens ? ` (in ${stats.inputTokens ?? 0}, out ${stats.outputTokens ?? 0})` : ''}`,
+      );
+    }
+    return lines.join('\n');
+  }
+
+  formatDetailed(taskDesc: string, now = Date.now()): string {
+    const stats = this.getSummary(now);
+    const sr =
+      stats.totalToolCalls > 0
+        ? (stats.successRate ??
+          (stats.successfulToolCalls / stats.totalToolCalls) * 100)
+        : 0;
+    const lines: string[] = [];
+    lines.push(`📋 Task Completed: ${taskDesc}`);
+    lines.push(
+      `⏱️ Duration: ${this.fmtDuration(stats.totalDurationMs)} | 🔁 Rounds: ${stats.rounds}`,
+    );
+    // Quality indicator
+    let quality = 'Poor execution';
+    if (sr >= 95) quality = 'Excellent execution';
+    else if (sr >= 85) quality = 'Good execution';
+    else if (sr >= 70) quality = 'Fair execution';
+    lines.push(`✅ Quality: ${quality} (${sr.toFixed(1)}% tool success)`);
+    // Speed category
+    const d = stats.totalDurationMs;
+    let speed = 'Long execution - consider breaking down tasks';
+    if (d < 10_000) speed = 'Fast completion - under 10 seconds';
+    else if (d < 60_000) speed = 'Good speed - under a minute';
+    else if (d < 300_000) speed = 'Moderate duration - a few minutes';
+    lines.push(`🚀 Speed: ${speed}`);
+    lines.push(
+      `🔧 Tools: ${stats.totalToolCalls} calls, ${sr.toFixed(1)}% success (${stats.successfulToolCalls} ok, ${stats.failedToolCalls} failed)`,
+    );
+    if (typeof stats.totalTokens === 'number') {
+      lines.push(
+        `🔢 Tokens: ${stats.totalTokens.toLocaleString()} (in ${stats.inputTokens ?? 0}, out ${stats.outputTokens ?? 0})`,
+      );
+    }
+    if (stats.toolUsage && stats.toolUsage.length) {
+      const sorted = [...stats.toolUsage]
+        .sort((a, b) => b.count - a.count)
+        .slice(0, 5);
+      lines.push('\nTop tools:');
+      for (const t of sorted) {
+        const avg =
+          typeof t.averageDurationMs === 'number'
+            ? `, avg ${this.fmtDuration(Math.round(t.averageDurationMs))}`
+            : '';
+        lines.push(
+          ` - ${t.name}: ${t.count} calls (${t.success} ok, ${t.failure} fail${avg}${t.lastError ? `, last error: ${t.lastError}` : ''})`,
+        );
+      }
+    }
+    const tips = this.generatePerformanceTips(stats);
+    if (tips.length) {
+      lines.push('\n💡 Performance Insights:');
+      for (const tip of tips.slice(0, 3)) lines.push(` - ${tip}`);
+    }
+    return lines.join('\n');
+  }
+
+  private fmtDuration(ms: number): string {
+    if (ms < 1000) return `${Math.round(ms)}ms`;
+    if (ms < 60000) return `${(ms / 1000).toFixed(1)}s`;
+    if (ms < 3600000) {
+      const m = Math.floor(ms / 60000);
+      const s = Math.floor((ms % 60000) / 1000);
+      return `${m}m ${s}s`;
+    }
+    const h = Math.floor(ms / 3600000);
+    const m = Math.floor((ms % 3600000) / 60000);
+    return `${h}h ${m}m`;
+  }
+
+  private generatePerformanceTips(stats: SubagentStatsSummary): string[] {
+    const tips: string[] = [];
+    const totalCalls = stats.totalToolCalls;
+    const sr =
+      stats.totalToolCalls > 0
+        ? (stats.successRate ??
+          (stats.successfulToolCalls / stats.totalToolCalls) * 100)
+        : 0;
+
+    // High failure rate
+    if (sr < 80)
+      tips.push('Low tool success rate - review inputs and error messages');
+
+    // Long duration
+    if (stats.totalDurationMs > 60_000)
+      tips.push('Long execution time - consider breaking down complex tasks');
+
+    // Token usage
+    if (typeof stats.totalTokens === 'number' && stats.totalTokens > 100_000) {
+      tips.push(
+        'High token usage - consider optimizing prompts or narrowing scope',
+      );
+    }
+    if (typeof stats.totalTokens === 'number' && totalCalls > 0) {
+      const avgTokPerCall = stats.totalTokens / totalCalls;
+      if (avgTokPerCall > 5_000)
+        tips.push(
+          `High token usage per tool call (~${Math.round(avgTokPerCall)} tokens/call)`,
+        );
+    }
+
+    // Network failures
+    const isNetworkTool = (name: string) => /web|fetch|search/i.test(name);
+    const hadNetworkFailure = (stats.toolUsage || []).some(
+      (t) =>
+        isNetworkTool(t.name) &&
+        t.lastError &&
+        /timeout|network/i.test(t.lastError),
+    );
+    if (hadNetworkFailure)
+      tips.push(
+        'Network operations had failures - consider increasing timeout or checking connectivity',
+      );
+
+    // Slow tools
+    const slow = (stats.toolUsage || [])
+      .filter((t) => (t.averageDurationMs ?? 0) > 10_000)
+      .sort((a, b) => (b.averageDurationMs ?? 0) - (a.averageDurationMs ?? 0));
+    if (slow.length)
+      tips.push(
+        `Consider optimizing ${slow[0].name} operations (avg ${this.fmtDuration(Math.round(slow[0].averageDurationMs!))})`,
+      );
+
+    return tips;
+  }
 }
--- a/packages/core/src/subagents/subagent.ts
+++ b/packages/core/src/subagents/subagent.ts
@@ -38,7 +38,6 @@ import {
  SubAgentStreamTextEvent,
  SubAgentErrorEvent,
 } from './subagent-events.js';
-import { formatCompact } from './subagent-result-format.js';
 import {
  SubagentStatistics,
  SubagentStatsSummary,
@@ -551,8 +550,7 @@ export class SubAgentScope {
        {
          terminate_reason: this.output.terminate_reason,
          result: this.finalText,
-          execution_summary: formatCompact(
-            summary,
+          execution_summary: this.stats.formatCompact(
            'Subagent execution completed',
          ),
        },
--- a/packages/core/src/subagents/types.ts
+++ b/packages/core/src/subagents/types.ts
@@ -10,8 +10,9 @@ import { Content, FunctionDeclaration } from '@google/genai';
 * Represents the storage level for a subagent configuration.
 * - 'project': Stored in `.qwen/agents/` within the project directory
 * - 'user': Stored in `~/.qwen/agents/` in the user's home directory
+ * - 'builtin': Built-in agents embedded in the codebase, always available
 */
-export type SubagentLevel = 'project' | 'user';
+export type SubagentLevel = 'project' | 'user' | 'builtin';

 /**
 * Core configuration for a subagent as stored in Markdown files.
@@ -60,6 +61,12 @@ export interface SubagentConfig {
   * If 'auto' or omitted, uses automatic color assignment.
   */
  color?: string;
+
+  /**
+   * Indicates whether this is a built-in agent.
+   * Built-in agents cannot be modified or deleted.
+   */
+  readonly isBuiltin?: boolean;
 }

 /**
--- a/packages/core/src/tools/task.test.ts
+++ b/packages/core/src/tools/task.test.ts
@@ -96,7 +96,7 @@ describe('TaskTool', () => {
    it('should initialize with correct name and properties', () => {
      expect(taskTool.name).toBe('task');
      expect(taskTool.displayName).toBe('Task');
-      expect(taskTool.kind).toBe('execute');
+      expect(taskTool.kind).toBe('other');
    });

    it('should load available subagents during initialization', () => {
--- a/packages/core/src/tools/task.ts
+++ b/packages/core/src/tools/task.ts
@@ -70,7 +70,7 @@ export class TaskTool extends BaseDeclarativeTool<TaskParams, ToolResult> {
      TaskTool.Name,
      'Task',
      'Delegate tasks to specialized subagents. Loading available subagents...', // Initial description
-      Kind.Execute,
+      Kind.Other,
      initialSchema,
      true, // isOutputMarkdown
      true, // canUpdateOutput - Enable live output updates for real-time progress