Show HN: Showboat and Rodney, so agents can demo what they've built
Simon Willison introduces Showboat and Rodney, two new CLI tools designed to help AI coding agents demonstrate their work and prove functionality. Showboat allows agents to construct Markdown-based demos with executed commands and screenshots, while Rodney provides browser automation for web-based projects. This development tackles the challenge of verifying agent-generated code beyond automated tests, reflecting HN's interest in practical applications of AI in software development.
The Lowdown
Simon Willison has released two new open-source tools, Showboat and Rodney, aimed at addressing a critical challenge in working with AI coding agents: how to effectively demonstrate and verify their work. While automated tests prove code correctness, they often don't fully illustrate functionality or user experience. These tools provide a structured way for agents to create compelling, artifact-rich demonstrations for human overseers.<ul><li><b>The Problem:</b> Current AI coding agents lack robust methods to demonstrate their built software beyond unit tests, leading to a need for human manual QA or expensive agent-based QA. The goal is to get agents to clearly show their work and minimize opportunities for "cheating."</li><li><b>Showboat:</b> This Go-based CLI tool enables AI agents to construct a Markdown document. Agents use commands like init, note, exec, and image to document steps, execute code, capture outputs, and embed screenshots directly into the demo document.</li><li><b>Agent-Centric Design:</b> Showboat is specifically designed for agents, using its comprehensive --help output as a self-contained guide for AI models like Claude to understand and utilize its full capabilities. This allows agents to create real-time, updating demo documents.</li><li><b>Rodney:</b> A CLI browser automation tool built on the Go-Rod library, Rodney allows agents to perform multi-turn browser sessions. It provides commands to start Chrome, open URLs, execute js, click elements, and take screenshots, integrating seamlessly with Showboat for web-based demonstrations.</li><li><b>Motivation & Workflow:</b> Willison emphasizes that while TDD is useful for agents, automated tests alone are insufficient. Showboat and Rodney bridge the gap by providing "manual" testing in an automated way, ensuring that functionality is visually confirmed.</li><li><b>Development Insight:</b> Notably, both tools were developed primarily by the author using Claude Code for web via his iPhone, highlighting the increasing viability of mobile-driven, AI-assisted coding workflows.</li></ul>By providing agents with the means to generate verifiable, visual demonstrations, Showboat and Rodney aim to enhance trust and efficiency in AI-assisted software development, ensuring human overseers can quickly grasp what the agents have built without needing to dive deep into the code or run extensive manual tests themselves.