What problem does it solve?
Programmatically posing or animating a humanoid figure requires anatomically correct joint limits, forward kinematics, and a renderer — this Skill provides a validated 22-bone skeleton rig with clinical range-of-motion clamps, a zero-dependency Node CLI, and a self-contained Three.js viewer that agents can also drive live over WebMCP.
Core Features & Use Cases
- Anatomical movement protocol: Compose pose commands using real movement names (flexion, abduction, pronation) that are validated against per-joint allowlists and clamped to cited range-of-motion limits, with swing-twist cone constraints for ball joints.
- Animation sequences: Validate and play keyframed walk/run/dance/backflip sequences with looped Catmull-Rom interpolation, grounding, and an optional root 6-DOF channel for travel, turns, and jumps.
- Agent-driven control: The generated viewer registers 11 WebMCP tools (apply_pose, reach, look_at, play_sequence, get_scheme) so an agent can drive the figure in joint space or task space over Chrome DevTools Protocol.
- Use Case: Ask the agent to animate a run cycle — it inspects the skeleton scheme, validates
assets/run-cycle.json with the sequence command, generates mannequin.html, and plays the looped animation in the browser.
Quick Start
Ask the agent to pose the manikin with the left arm flexed 90 degrees and render it to an HTML viewer using the octocode-mannequin skill.