What is a Skill?
SUMMARY
A Skill is a self-contained, reusable operation with strictly typed inputs and outputs - the fundamental building block of a Telekinesis application.
Skills are to Physical AI what functions are to software: small, composable units that combine into larger programs. Physical AI Agents build those programs by selecting and parameterizing Skills.
In the Telekinesis ecosystem, a Skill is the fundamental building block for Physical AI applications.
A Skill is a self-contained, reusable operation that performs a specific task such as 3D perception, image processing, motion planning, or decision logic.
A useful mental model: Skills are to Physical AI what functions are to software - small, composable units that can be combined into larger programs.
Skill Interface
Each Skill is defined as a strongly typed function-like interface:
outputs = skill(inputs)Where:
- Inputs: strict data types defined in the Data Engine Layer
- Outputs: strict data types defined in the Data Engine Layer
This allows for seamless composition, as the output of one Skill can be directly fed into another.
Example Skill
from telekinesis import vitreous
# Beginner-friendly usage
pc = vitreous.load_point_cloud('path/to/point_cloud.ply')
downsampled_pc = vitreous.filter_point_cloud_using_voxel_downsampling(
point_cloud=pc,
voxel_size=0.01
)How Skills Are Organized
Skills are grouped into modules by domain. Each module is imported directly from the telekinesis library:
from telekinesis import synapse # robotics: kinematics, planning, control, communication
from telekinesis import cornea # image segmentation
from telekinesis import retina # object detection
from telekinesis import pupil # image processing
from telekinesis import vitreous # 3D point cloud processing
from telekinesis import medulla # hardware communication
from telekinesis import iris # AI model trainingRobotics Skills
Full robotics stack for motion planning, kinematics, control, and robot and tool communication supporting Universal Robots, FANUC, KUKA, Robotiq, OnRobot and more.
Explore →
Image Segmentation Skills
2D image segmentation from color spaces (RGB, HSV, LAB, YCrCb) to deep learning (SAM, BiRefNet), with GrabCut, watershed, SLIC, Felzenszwalb, Otsu, and adaptive thresholding.
Explore →
Object Detection Skills
2D object detection from classical shapes (Hough, contours) to deep learning (YOLOX, RF-DETR), with open-vocabulary models (Grounding DINO, Qwen-VL) for zero-shot recognition.
Explore →3D Point Cloud Processing Skills
3D point cloud processing from filtering (voxel, outlier removal) to clustering (DBSCAN, density-jump), registration (ICP, FPFH), and mesh reconstruction (Poisson, convex hull).
Explore →
Image Processing Skills
Low-level image processing from morphology (erode, dilate, top-hat) to filtering (Gaussian, bilateral, CLAHE), edge detection (Sobel, Frangi, Gabor), and geometric transforms.
Explore →
Hardware Communication Skills
Cameras, sensors, and other hardware communication for robot perception pipelines, with streaming for color, depth, and point cloud data from 2D and 3D cameras.
Explore →
AI Model Training Skills
Flexible framework for managing datasets, training pipelines, and model deployment.
Explore →Reinforcement Learning Skills
Reinforcement learning primitives for robot control, sim-to-real transfer, and policy deployment.
Explore →How Skills Enable Physical AI Agents
In the Telekinesis ecosystem, Physical AI Agents generate programs by selecting, composing, and parameterizing Skills from the Telekinesis Agentic Skill Library. Instead of directly outputting robot actions, the agent produces structured Python code that orchestrates Skills into executable workflows.
This creates a clear separation between:
- Reasoning (Agent layer) - decides what to do
- Execution (Skill layer) - defines how it is done
A Physical AI workflow typically follows this structure:
User Instruction
↓
Agent reasoning (LLM / VLM)
↓
Skill selection (from library)
↓
Skill composition (Python program)
↓
Execution in standard Python runtimeThe diagram below illustrates how Physical AI Agents interact with the Skill Library to generate executable robotics programs.
