Workatron turns any machine into a tentacle node that runs your pipelines, manages resource-hungry processors, and streams live results back to a single mothership dashboard — all from one self-contained binary.
Workatron bundles orchestration, process management, secure federation, and a live UI into a single deployable unit — no brokers, agents, or databases to stand up first.
Compose multi-step pipelines from reusable steps — chain executables, fan out with foreach, declare requires dependencies, run steps concurrently, and bound each with a timeout. Parameters are templated and validated before a job ever runs.
Processor units wrap your executables or Python virtualenvs as long-lived workers. Schedule them against a weight budget — VRAM, RAM, anything you define — with readiness probes, idle shutdown, and automatic restart-with-backoff.
A mothership aggregates many tentacle nodes over secure, resumable, multiplexed RPC connections that survive network drops. Target a single node or fan one job across the whole fleet.
The dashboard subscribes to job and node state over Server-Sent Events, transported as a compact diff/patch stream — only what actually changed crosses the wire, so thousands of jobs stay smooth.
Submit jobs with uploaded file inputs, idempotency keys, and priorities. Watch each step advance through its lifecycle, then pull outputs back one file at a time or as a single zip.
Push a signed update package and the fleet handles the rest — every node verifies the publisher's public key, swaps its own binary, and relaunches. No external orchestration required.
A single trimmed, self-extracting executable embeds both the HTTP API and the Angular UI. Run it in the foreground or install it as a Windows Service or systemd daemon with one flag.
Guard the API with HTTPS and a bearer token; node-to-node links authenticate both ends with public-key-pinned certificates. Retention windows age out finished and failed jobs automatically.
Every node is a complete Workatron on its own. Promote one to a mothership and it federates the rest over secure, resumable, multiplexed RPC — then any operation can target a single node or fan out across all of them.
Tentacle node
Tentacle node
Tentacle node
Tentacle nodeJobs move through a small, predictable set of states. The dashboard reflects each transition live, and finished or failed jobs age out automatically on the retention window you configure.
Workatron ships as a single executable named workatron. Point it at a workspace folder and it does the rest.
Launch workatron tentacle on any machine. It writes a starter config.json and comes up ready to work.
Register processor units and compose pipelines from the dashboard, or drop JSON straight into the workspace folder.
POST a job with its inputs to a group. Workatron schedules the steps against available resources and tracks every transition.
Turn on the mothership to federate more tentacle nodes, then target one or fan jobs across the entire fleet.
# Start a tentacle node (workspace defaults to %APPDATA%\workatron)
$ workatron tentacle
# Use an explicit workspace directory
$ workatron tentacle /srv/workatron
# Install it as a background service (run elevated)
$ workatron tentacle --install --name workatron
# Remove the service again
$ workatron tentacle --uninstall --name workatronFirst run writes config.json. Flip on the HTTPS server to expose the API and the embedded dashboard, then secure it with a bearer token.