SpokeUI: Point, Speak, Ship UI Changes

SpokeUI is a local, voice-driven visual coding workspace that removes the translation step between seeing a UI problem and explaining it to a coding agent. Instead of finding a component name, locating its source file, taking screenshots, or writing a long prompt, the developer interacts directly with the running application. Hover over the interface, select the exact element you want to change, hold Space, and describe the desired result naturally. SpokeUI captures the selected element's DOM ancestry, styles, geometry, viewport and relevant runtime context. AssemblyAI provides real-time streaming speech-to-text, converting the developer's spoken intent into an actionable instruction. That context is then passed to an authenticated local Codex or Claude Code runtime, which works against the real repository and modifies the actual source files. The running preview updates with the result, and SpokeUI presents before-and-after captures, changed-file information, and an Accept or Reject workflow so the developer stays in control. SpokeUI also supports debugging directly from the rendered interface by combining selected DOM context with console and network evidence. The idea is simple: developers already know what looks wrong when they see it. They should not have to explain where it exists in the codebase before an AI coding agent can help.