research_task
Perform targeted web research, generate intelligent suggestions, and store findings in memory for efficient task management. Combines web research with local knowledge caching to enhance research workflow.
Instructions
Guide the AI agent to perform comprehensive web research for a task, with intelligent research suggestions and automatic memory storage of findings. Combines web research capabilities with local knowledge caching for optimal research workflow.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| checkExistingMemories | No | Whether to check existing memories first | |
| researchAreas | No | Specific areas to research (auto-generated if not provided) | |
| researchDepth | No | Depth of research to perform | standard |
| saveToMemories | No | Whether to save research findings to memories | |
| taskId | Yes | ID of the task to research | |
| workingDirectory | Yes | The full absolute path to the working directory where data is stored. MUST be an absolute path, never relative. Windows: "C:\Users\username\project" or "D:\projects\my-app". Unix/Linux/macOS: "/home/username/project" or "/Users/username/project". Do NOT use: ".", "..", "~", "./folder", "../folder" or any relative paths. Ensure the path exists and is accessible before calling this tool. NOTE: When server is started with --claude flag, this parameter is ignored and a global user directory is used instead. |
Implementation Reference
- The main execution handler for the 'research_task' tool. It retrieves the task, checks existing memories, generates research areas and guidance, builds a comprehensive response with instructions for the AI agent, and handles errors.handler: async (args: any) => { try { const { workingDirectory, taskId, researchAreas, saveToMemories, checkExistingMemories, researchDepth } = args; // Get the task to research const task = await storage.getTask(taskId); if (!task) { return { content: [{ type: 'text' as const, text: `Error: Task with ID "${taskId}" not found.` }], isError: true }; } // Check existing memories first if requested let existingKnowledge = ''; if (checkExistingMemories) { existingKnowledge = await checkExistingMemoriesForTask(memoryStorage, task); } // Generate research areas if not provided const finalResearchAreas = researchAreas || generateResearchAreas(task); // Generate research guidance for the AI agent const researchGuidance = generateResearchGuidance( task, finalResearchAreas, researchDepth, existingKnowledge ); // Create memory storage instructions if requested const memoryInstructions = saveToMemories ? generateMemoryStorageInstructions(task, finalResearchAreas) : ''; // Build the response text safely let responseText = `🔍 **Task Research Guidance for AI Agent** **Task to Research:** ${task.name} **Task Details:** ${task.details} **Priority:** ${task.priority}/10 **Complexity:** ${task.complexity || 'Not set'}/10 **Tags:** ${task.tags?.join(', ') || 'None'}`; if (existingKnowledge) { responseText += `\n\n📚 **Existing Knowledge Found:**\n${existingKnowledge}`; } responseText += `\n\n🎯 **Research Areas to Investigate:**\n`; responseText += finalResearchAreas.map((area: string, index: number) => `${index + 1}. ${area}`).join('\n'); responseText += `\n\n${researchGuidance}`; if (memoryInstructions) { responseText += `\n\n${memoryInstructions}`; } responseText += `\n\n👉 **Your Actions: Post-Research Task Updates & Next Steps** Once your research is complete and findings are stored: 1. **Update Task Details:** If your research uncovered new information, requirements, or changes to the task's scope, duration, or complexity, use the \`update_task\` tool. * Example: \`update_task({ id: "${task.id}", details: "new details based on research", complexity: new_complexity_value })\` 2. **Analyze Complexity & Breakdown (If Needed):** If the research suggests the task is significantly more complex than initially thought, it might need to be broken down. * You can use \`analyze_task_complexity\` to get suggestions: \`analyze_task_complexity({ taskId: "${task.id}" })\` * Alternatively, create nested tasks using \`create_task\` with parentId: \`create_task({ projectId: "${task.projectId}", parentId: "${task.id}", name: "subtask_name", details: "..." })\` 3. **Determine Next Task:** To see what task (including potentially this one, if it's now unblocked or its priority has increased) is best to work on next, use the \`get_next_task_recommendation\` tool. * Example: \`get_next_task_recommendation({ projectId: "${task.projectId}" })\``; return { content: [{ type: 'text' as const, text: responseText }] }; } catch (error: any) { return { content: [{ type: 'text' as const, text: `Error generating research guidance: ${error instanceof Error ? error.message : 'Unknown error'}` }], isError: true }; }
- Zod input schema defining parameters for the research_task tool: workingDirectory, taskId, optional researchAreas, saveToMemories, checkExistingMemories, and researchDepth.inputSchema: z.object({ workingDirectory: z.string().describe(getWorkingDirectoryDescription(config)), taskId: z.string().describe('ID of the task to research'), researchAreas: z.array(z.string()).optional().describe('Specific areas to research (auto-generated if not provided)'), saveToMemories: z.boolean().optional().default(true).describe('Whether to save research findings to memories'), checkExistingMemories: z.boolean().optional().default(true).describe('Whether to check existing memories first'), researchDepth: z.enum(['quick', 'standard', 'comprehensive']).optional().default('standard').describe('Depth of research to perform') }),
- Factory function that creates and returns the complete tool object for 'research_task', including name, description, inputSchema, and handler. This is the primary definition point for the tool.export function createTaskResearchTool( storage: Storage, memoryStorage: MemoryFileStorage, getWorkingDirectoryDescription: (config: any) => string, config: any ) { return { name: 'research_task', description: 'Guide the AI agent to perform comprehensive web research for a task, with intelligent research suggestions and automatic memory storage of findings. Combines web research capabilities with local knowledge caching.', inputSchema: z.object({ workingDirectory: z.string().describe(getWorkingDirectoryDescription(config)), taskId: z.string().describe('ID of the task to research'), researchAreas: z.array(z.string()).optional().describe('Specific areas to research (auto-generated if not provided)'), saveToMemories: z.boolean().optional().default(true).describe('Whether to save research findings to memories'), checkExistingMemories: z.boolean().optional().default(true).describe('Whether to check existing memories first'), researchDepth: z.enum(['quick', 'standard', 'comprehensive']).optional().default('standard').describe('Depth of research to perform') }), handler: async (args: any) => { try { const { workingDirectory, taskId, researchAreas, saveToMemories, checkExistingMemories, researchDepth } = args; // Get the task to research const task = await storage.getTask(taskId); if (!task) { return { content: [{ type: 'text' as const, text: `Error: Task with ID "${taskId}" not found.` }], isError: true }; } // Check existing memories first if requested let existingKnowledge = ''; if (checkExistingMemories) { existingKnowledge = await checkExistingMemoriesForTask(memoryStorage, task); } // Generate research areas if not provided const finalResearchAreas = researchAreas || generateResearchAreas(task); // Generate research guidance for the AI agent const researchGuidance = generateResearchGuidance( task, finalResearchAreas, researchDepth, existingKnowledge ); // Create memory storage instructions if requested const memoryInstructions = saveToMemories ? generateMemoryStorageInstructions(task, finalResearchAreas) : ''; // Build the response text safely let responseText = `🔍 **Task Research Guidance for AI Agent** **Task to Research:** ${task.name} **Task Details:** ${task.details} **Priority:** ${task.priority}/10 **Complexity:** ${task.complexity || 'Not set'}/10 **Tags:** ${task.tags?.join(', ') || 'None'}`; if (existingKnowledge) { responseText += `\n\n📚 **Existing Knowledge Found:**\n${existingKnowledge}`; } responseText += `\n\n🎯 **Research Areas to Investigate:**\n`; responseText += finalResearchAreas.map((area: string, index: number) => `${index + 1}. ${area}`).join('\n'); responseText += `\n\n${researchGuidance}`; if (memoryInstructions) { responseText += `\n\n${memoryInstructions}`; } responseText += `\n\n👉 **Your Actions: Post-Research Task Updates & Next Steps** Once your research is complete and findings are stored: 1. **Update Task Details:** If your research uncovered new information, requirements, or changes to the task's scope, duration, or complexity, use the \`update_task\` tool. * Example: \`update_task({ id: "${task.id}", details: "new details based on research", complexity: new_complexity_value })\` 2. **Analyze Complexity & Breakdown (If Needed):** If the research suggests the task is significantly more complex than initially thought, it might need to be broken down. * You can use \`analyze_task_complexity\` to get suggestions: \`analyze_task_complexity({ taskId: "${task.id}" })\` * Alternatively, create nested tasks using \`create_task\` with parentId: \`create_task({ projectId: "${task.projectId}", parentId: "${task.id}", name: "subtask_name", details: "..." })\` 3. **Determine Next Task:** To see what task (including potentially this one, if it's now unblocked or its priority has increased) is best to work on next, use the \`get_next_task_recommendation\` tool. * Example: \`get_next_task_recommendation({ projectId: "${task.projectId}" })\``; return { content: [{ type: 'text' as const, text: responseText }] }; } catch (error: any) { return { content: [{ type: 'text' as const, text: `Error generating research guidance: ${error instanceof Error ? error.message : 'Unknown error'}` }], isError: true }; } } }; }
- Helper function that intelligently generates specific research areas from task content, detecting technologies, tags, and complexity to suggest relevant investigation topics.function generateResearchAreas(task: Task): string[] { const areas: string[] = []; const content = (task.name + ' ' + task.details).toLowerCase(); // Technology-specific research areas const techPatterns = { 'authentication': ['OAuth2 best practices', 'JWT security considerations', 'Multi-factor authentication implementation'], 'database': ['Database schema design', 'Performance optimization techniques', 'Data migration strategies'], 'api': ['REST API design principles', 'API security best practices', 'Rate limiting strategies'], 'frontend': ['Modern UI/UX patterns', 'Accessibility guidelines', 'Performance optimization'], 'testing': ['Testing strategies and frameworks', 'Test automation best practices', 'Coverage requirements'], 'deployment': ['CI/CD pipeline setup', 'Container orchestration', 'Monitoring and logging'], 'security': ['Security vulnerability assessment', 'Encryption standards', 'Compliance requirements'], 'performance': ['Performance benchmarking', 'Optimization techniques', 'Scalability patterns'] }; // Check for technology keywords and add relevant research areas for (const [tech, researchTopics] of Object.entries(techPatterns)) { if (content.includes(tech)) { areas.push(...researchTopics); } } // Add general research areas based on task tags if (task.tags) { for (const tag of task.tags) { areas.push(`${tag} best practices and current trends`); areas.push(`Common pitfalls and solutions for ${tag}`); } } // Add complexity-based research if (task.complexity && task.complexity >= 7) { areas.push('Architecture patterns for complex implementations'); areas.push('Risk mitigation strategies'); } // Default research areas if none found if (areas.length === 0) { areas.push(`Best practices for: ${task.name}`); areas.push('Implementation approaches and alternatives'); areas.push('Common challenges and solutions'); areas.push('Required tools and dependencies'); } // Remove duplicates and limit to reasonable number return [...new Set(areas)].slice(0, 8); }
- Helper function that generates comprehensive, depth-specific research guidance and instructions for the AI agent to follow during web research.function generateResearchGuidance( task: Task, researchAreas: string[], depth: string, existingKnowledge: string ): string { const depthGuidance = { quick: { timeframe: '15-30 minutes', focus: 'Key concepts and immediate implementation needs', sources: '2-3 authoritative sources per area' }, standard: { timeframe: '30-60 minutes', focus: 'Comprehensive understanding and implementation details', sources: '3-5 diverse sources per area including documentation, tutorials, and best practices' }, comprehensive: { timeframe: '1-2 hours', focus: 'Deep understanding, alternatives, edge cases, and long-term considerations', sources: '5+ sources including academic papers, case studies, and expert opinions' } }; const guidance = depthGuidance[depth as keyof typeof depthGuidance]; return `👉 **Your Action: Perform Web Research** Now, use your web search capabilities to investigate each of the "Research Areas to Investigate" listed above. Follow the strategy outlined below. If you need help formulating specific search queries for this task, you can use the \`generate_research_queries\` tool. Example: \`generate_research_queries({ taskId: "${task.id}" })\` 📋 **Research Instructions for AI Agent:** **Research Depth:** ${depth.toUpperCase()} (${guidance.timeframe}) **Focus:** ${guidance.focus} **Sources per area:** ${guidance.sources} 🌐 **Web Research Strategy:** 1. **Start with official documentation** for any technologies mentioned 2. **Look for recent tutorials and guides** (prefer content from last 2 years) 3. **Check Stack Overflow and GitHub** for real-world implementation examples 4. **Review best practices articles** from reputable tech blogs 5. **Look for case studies** of similar implementations ${depth === 'comprehensive' ? '6. **Search for academic papers** or research on the topic\n7. **Find expert opinions** and industry reports' : ''} 🔍 **For Each Research Area:** - Search for: "[area] + implementation guide + ${new Date().getFullYear()}" - Look for: "[area] + best practices + examples" - Check: "[area] + common mistakes + solutions" - Find: "[area] + performance + optimization" ⚠️ **Research Quality Criteria:** - Prioritize sources from last 2 years for technology topics - Verify information across multiple sources - Look for code examples and practical implementations - Note any version-specific considerations - Identify potential compatibility issues 🎯 **Expected Research Outcomes:** - Clear understanding of implementation approach - List of required tools/dependencies - Identification of potential challenges - Time estimate refinement based on research - Alternative approaches if primary approach has issues`; }