Charakuru packages an agent-assisted route from illustration to VRM
Charakuru is a local application and workflow kit for turning a single full-body character illustration into a VRM avatar. Its README says the workflow uses Tripo’s Smart Mesh and delegates image preparation, Tripo and Blender operations, rigging and export to an AI agent, while the user reviews and directs the work in the app.
The workflow is explicitly checkpointed: users review generated images, choose among 3D candidates, adjust proportions and inspect motion. The agent pauses at those decision points rather than selecting candidates or advancing autonomously. A project is kept in one project file, and restarting the app from its folder reopens the work.
The repository describes the kit as still in development and lists requirements including Node.js 22 or later, Codex with image generation and browser control, a Tripo account, Blender 5.2, a VRM add-on and Blender MCP. External services are used with the user’s own accounts. The README directs users to a release ZIP for the runnable app; GitHub’s automatic source archive contains documentation and images, not the runnable kit.