Agent Roles & Specialization

Master specialized agent roles and build effective multi-agent teams

The Four Core Agent Roles

Most multi-agent systems are built around four foundational roles. Each role addresses a distinct phase of problem-solving, creating a complete research → plan → execute → review workflow.

The RPEC Framework

🔍
Researcher: Gathers information
📋
Planner: Designs strategy
Executor: Takes action
Critic: Ensures quality

Interactive: Role Evolution Simulator

Select a role and adjust the evolution stage to see how each role becomes more sophisticated as task complexity increases.

🔍

Researcher

Information Gathering

CAPABILITY

Basic web search

COMPLEXITY LEVEL

AVAILABLE TOOLS

Web searchDatabase queries

Role Characteristics Comparison

RolePrimary OutputSuccess MetricFailure Mode
🔍 ResearcherInformation summaryRelevance & accuracyMissing data
📋 PlannerTask breakdownCompleteness & feasibilityUnrealistic plan
⚡ ExecutorImplemented solutionSuccess rate & speedIncorrect execution
✅ CriticQuality assessmentError detection rateFalse positives

💡 Key Insight

The four core roles form a complete feedback loop. The Critic's output informs the Researcher about gaps, which leads to better planning, which enables more effective execution. This cyclical improvement is why specialized roles outperform generalist approaches.