Documentation
•
2 min read

Agent Roles

Understanding the Architect, Coder, Reviewer, and Judge agents.

Last updated2026-02-28

AI Agent Flow uses a multi-agent "Chain of Thought" approach to ensure high-quality, reliable output. Each agent has a specific specialized role in the process.

Developer Insight: This multi-agent separation prevents "hallucination cascade" — where a single model might ignore its own errors. Each agent acts as a check on the previous one.

🏛️ The Architect

The Architect is the first agent to act. Its job is to:

  • Analyze the user request.
  • Read relevant files from the codebase.
  • Draft a step-by-step implementation plan.
  • Identify potential edge cases.

💻 The Coder

The Coder follows the Architect's plan. It:

  • Writes the actual code changes.
  • Ensures syntax correctness.
  • Implements the logic specified in the plan.

🔍 The Reviewer

The Reviewer acts as a second pair of eyes. It looks for:

  • Logic bugs.
  • Security vulnerabilities.
  • Style guide violations.
  • Inefficiencies or anti-patterns.

⚖️ The Judge

The Judge performs final validation. It:

  • Runs tests (if available).
  • Cross-references the output with the original user request.
  • Decides whether to PASS the task or send it back for another iteration.
NextProvider Configuration