Autonomous agent for research and analysis tasks

Launches a smart agent to handle complex searches and investigations. Great for large-scale work without eating up context.

Usage

bash
# Request Task from Claude
"Investigate [task] using Task"

What Task Does

Works Independently

• Combines multiple tools automatically
• Gathers and analyzes step by step
• Puts results together in clear reports

Saves Context

• Uses less memory than manual searching
• Searches lots of files efficiently
• Pulls data from outside sources

Ensures Quality

• Checks if sources are reliable
• Verifies from different angles
• Fills in missing pieces

Basic Examples

bash
# Complex codebase investigation
"Investigate which files implement this feature using Task"

# Large-scale file search
"Identify configuration file inconsistencies using Task"

# External information collection
"Investigate the latest AI technology trends using Task"

Collaboration with Claude

bash
# Complex problem analysis
"Analyze the cause of memory leaks using Task, including profiling results and logs"

# Dependency investigation
"Investigate vulnerabilities of this npm package using Task"

# Competitor analysis
"Investigate API specifications of competing services using Task"

# Architecture analysis
"Analyze dependencies of this microservice using Task"

Task vs Other Commands

When to Use What

| Command | Main Use Case | Execution Method | Information Collection | | ------------------- | ------------------------------- | --------------------- | -------------------------- | | Task | Investigation, analysis, search | Autonomous execution | Multiple sources | | ultrathink | Deep thinking, judgment | Structured thinking | Existing knowledge-focused | | sequential-thinking | Problem-solving, design | Step-by-step thinking | As needed | | plan | Implementation planning | Approval process | Requirement analysis |

Quick Decision Guide

text
Need to gather info?
├─ Yes → From many places or lots of files?
│          ├─ Yes → **Use Task**
│          └─ No → Just ask normally
└─ No → Need deep thinking?
          ├─ Yes → Use ultrathink/sequential-thinking
          └─ No → Just ask normally

When Task Works Best

Great For

• Exploring complex codebases (dependencies, architecture)
• Searching many files (patterns, configs)
• Gathering external info (tech trends, libraries)
• Combining data from multiple places (logs, metrics)
• Repetitive investigations (audits, debt checks)
• Big searches that would eat too much context

Not Great For

• Simple questions I already know
• Quick one-time tasks
• Things needing back-and-forth discussion
• Design decisions (use plan or thinking commands instead)

Detailed Examples by Category

System Analysis and Investigation

bash
# Complex system analysis
"Identify bottlenecks in the EC site using Task, investigating database, API, and frontend"

# Architecture analysis
"Analyze dependencies of this microservice using Task, including API communication and data flow"

# Technical debt investigation
"Analyze technical debt in legacy code using Task, including refactoring priorities"

Security and Compliance

bash
# Security audit
"Investigate vulnerabilities in this application using Task, based on OWASP Top 10"

# License investigation
"Investigate license issues in project dependencies using Task"

# Configuration file audit
"Identify security configuration inconsistencies using Task, including environment differences"

Performance and Optimization

bash
# Performance analysis
"Identify heavy queries in the application using Task, including execution plans and optimization proposals"

# Resource usage investigation
"Investigate causes of memory leaks using Task, including profiling results and code analysis"

# Bundle size analysis
"Investigate frontend bundle size issues using Task, including optimization suggestions"

External Information Collection

bash
# Technology trend investigation
"Investigate 2026 JavaScript framework trends using Task"

# Competitor analysis
"Investigate API specifications of competing services using Task, including feature comparison table"

# Library evaluation
"Compare state management libraries using Task, including performance and learning costs"

Execution Flow and Quality Assurance

Task Execution Flow

text
1. Initial Analysis
   ├─ Decomposition of task and identification of investigation scope
   ├─ Selection of necessary tools and information sources
   └─ Development of execution plan

2. Information Collection
   ├─ File search and code analysis
   ├─ Collection of external information
   └─ Data structuring

3. Analysis and Integration
   ├─ Relevance analysis of collected information
   ├─ Identification of patterns and issues
   └─ Verification of hypotheses

4. Reporting and Proposal
   ├─ Structuring of results
   ├─ Creation of improvement proposals
   └─ Presentation of next actions

Quality Assurance

• Reliability check of information sources: Fact confirmation from multiple sources
• Completeness check: Verification of no gaps in investigation targets
• Consistency verification: Confirmation of consistency in conflicting information
• Practicality evaluation: Assessment of feasibility and effectiveness of proposals

Error Handling and Constraints

Common Constraints

• External API usage limits: Rate limits and authentication errors
• Large file processing limits: Memory and timeout constraints
• Access permission issues: Restrictions on file and directory access

Error Handling

• Partial result reporting: Analysis with only obtainable information
• Alternative proposals: Suggestion of alternative investigation methods under constraints
• Stepwise execution: Division of large-scale tasks for execution

Notes

• Task is optimal for complex, autonomous investigation and analysis tasks
• For simple questions or when immediate answers are needed, use normal question format
• Treat investigation results as reference information and always verify important decisions
• When collecting external information, pay attention to the freshness and accuracy of information

Execution Example

bash
# Usage example
"Investigate issues in GraphQL schema using Task"

# Expected behavior
# 1. Dedicated agent starts
# 2. Search for GraphQL-related files
# 3. Analyze schema definitions
# 4. Compare with best practices
# 5. Identify issues and propose improvements
# 6. Create structured report