Blacknode Robotics
Visual runtime for intelligent systems

Build workflows.
Move machines.

Blacknode connects AI, GPU computing, automation, and robotics in one typed visual runtime—built for developers and AI agents to inspect, run, replay, and operate together.

Typed graphsLive runtimeSafe controlCode export
robot_cube_follow.workflow
LIVE
Camera
USB video
Object Tracker
Detection
green cube · 0.94
Robot
SO-ARM101
driver ready
Motion Dashboard
Live state
PERCEPTION
CONTROL
Hardware connected
60 Hz control loop
One runtime, many systems

Complex systems become
visible workflows.

Build from reusable nodes, see every connection, and keep execution state available for developers, operators, and agents.

01

Typed visual graphs

Compose data, tools, APIs, models, Python, GPU operations, and robot capabilities through validated ports.

ValidateOrganizeReuse
02

Inspectable runtime

Run workflows, watch live state, inspect failures, and replay every execution event.

03

Agent control surface

Give AI agents structured access through MCP, HTTP, and WebSocket APIs.

04

Safe robotics lifecycle

Discover hardware, monitor joints, gate motion, manage services, and stop explicitly.

05

Portable workflow artifacts

Save durable graphs and export readable Python or framework-ready code for deployment, sharing, and iteration.

$ blacknode export-python workflow.json
The robotics lifecycle

From first connection
to operational intelligence.

01

Build

Connect hardware, perception, models, and control.

02

Train

Record datasets and train robot policies.

03

Deploy

Move tested workflows into real environments.

04

Operate

Monitor state, replay runs, and manage services.

05

Scale

Grow repeatable workflows across a fleet.

Modular by design

A growing robotics
workflow ecosystem.

Blacknode packages remain independently installable while working together through stable typed contracts.

R

Robot

USB discovery, profiles, drivers, calibration, lifecycle.

blacknode-robot
R2

ROS 2

Topics, services, transport, control, dashboards.

blacknode-ros2
V

Vision

Cameras, OpenCV tracking, VLM reasoning, streams.

blacknode-vision
C

CUDA

GPU kernels, image filters, tensor operations.

blacknode-cuda
D

Dataset

Episode recording, validation, recovery, export.

blacknode-dataset
T

Training

Managed policy training, checkpoints, previews.

blacknode-training
Built for collaboration

Developers see the graph.
Agents use the same graph.

Blacknode gives AI agents a structured workflow surface while keeping nodes, connections, runtime events, and artifacts visible to the people building the system.

Read the MCP quickstart
blacknode — mcp

$ blacknode mcp

Blacknode MCP server ready

list node types

✓ 128 typed nodes available

build and validate workflow

✓ workflow valid · 9 nodes · 10 edges

run output.value

✓ run complete · replay saved

Open source · Apache 2.0

Start building with Blacknode.

Clone the repository, run one launcher, and open the visual editor.

git clone https://github.com/temiroff/Blacknode.git
cd Blacknode
.\start.bat