Introduction

GemKit agents are not just general-purpose AI; they are specialized personas with specific roles, instructions, and toolsets.

🔍 Researcher Agent

The Researcher is designed for deep technical investigation. It doesn't just give an answer; it validates information across multiple sources and synthesizes a structured report.

Best for: Learning new technologies, comparing frameworks, investigating bugs with unknown causes.
Key Skills: research, search-engine.
Command: gk agent spawn --agent researcher --prompt "Research..."

💻 Code-Executor Agent

The Code-Executor is the workhorse of the kit. It follows a plan-driven, test-driven approach to implementation.

Best for: Feature development, refactoring, bug fixing.
Key Skills: frontend-development, backend-development, filesystem.
Command: gk agent spawn --agent code-executor --prompt "Implement..."

🧪 Tester Agent

The Tester focuses on quality assurance. It analyzes your code and generates comprehensive test suites.

Best for: Unit testing, integration testing, increasing code coverage.
Key Skills: testing, jest, vitest, playwright.
Command: gk agent spawn --agent tester --prompt "Test this component..."

Agent Interactions

In GemKit, agents can spawn other agents. For example, the Code-Executor might spawn a Researcher to look up an API specification before implementing a feature, or a Tester to verify the changes it just made.