What problem does it solve? Reading a CARLA actor's state through separate per-actor getter calls can straddle two simulation frames, producing inconsistent position and velocity values. This Skill reads all kinematics from a single world snapshot and resolves ambiguous actor selections safely. ## Core Features & Use Cases - Frame-consistent snapshots: Reads transform, velocity, acceleration, and angular velocity from one world.get_snapshot() so all values belong to the same frame. - Vehicle-specific state: Reports control input (throttle/steer/brake), front-wheel steer angle, and mass for vehicles. - Flexible actor targeting: Select actors by id, role, type filter, color, or nearest-to-reference, with deliberate refusal of ambiguous matches. - Use Case: When a user asks "how fast is the ego going", run show --role hero to get speed in km/h plus the full state, or use watch to stream telemetry at a chosen rate. ## Quick Start Ask the agent to show the telemetry of the hero vehicle in the running CARLA server, or to stream a specific actor's speed for ten seconds.