Skip to main content

UnitPort local helper — headless daemon bridging local Isaac Lab / MuJoCo / resources to the UnitPort cloud canvas.

Project description

unitport — UnitPort local helper

Headless daemon that connects a local GPU / Isaac Lab host (workstation or the user's own cloud server) to the UnitPort cloud canvas. The canvas is the remote control (遥控器); this process has no GUI — only terminal output.

Authoritative design: ../knowledges/local_helper_design.md (rulings LH-D1…LH-D6, contracts C10 pairing / C11 control protocol).

What it is (and is not)

  • Is: an outbound-only WebSocket agent exposing a fixed verb set to the cloud (report capabilities, publish an asset index, receive a bundle from the user's own canvas, start/stop a training run, stream telemetry).
  • Is not: a shell, an exec server, or a GUI. The cloud can never send an arbitrary command — only the fixed verbs, only from a paired session of the same account, only bundles that originate from the user's own canvas (least-privilege red line, LH-D4 / C11).

The same binary serves execution target ① (UserServerTarget, user cloud server) and ③ (LocalHelperTarget, local host) — only the host differs (LH-D5).

Install

pip install unitport

Cross-platform (Windows + Linux). Runtime deps: websockets (the live link) and, on Python < 3.11, tomli (TOML reader). Everything else is stdlib.

CLI (headless — everything else is driven remotely from the canvas)

Command Purpose
unitport login Browser-approve device pairing; stores a refreshable helper token
unitport up (serve) Dial out to the cloud relay and serve the fixed verb set
unitport status Connection / run state, paired account, config summary
unitport config Declare Isaac/MuJoCo interpreter paths + resource roots
unitport doctor Fail-loud self-check (interpreters, deps, cloud, token)

There are deliberately no pull / train / push local subcommands — those are remote verbs, triggered from the canvas (C11 §4).

Config

  • Config dir: %APPDATA%\unitport\ (Windows) · $XDG_CONFIG_HOME/unitport/ or ~/.config/unitport/ (POSIX). Override with UNITPORT_HOME.
  • config.toml — cloud base URL, engine interpreter paths, resource_roots, workspace dir. Human-editable; also managed by unitport config.
  • token.json — the helper token (secret; chmod 600 on POSIX). Never printed, never logged.
  • device.json — the persistent (non-secret) device_id.

Status

Under active construction (phases LH-P1…LH-P8, Isaac Lab first). See the design doc §12 for the phase map. This tree currently implements the local foundation (config, capability detection, doctor, CLI) plus the pairing/transport clients; the workers/relay Durable Object cloud side lives in ../workers/relay.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

unitport-0.0.4.tar.gz (42.7 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

unitport-0.0.4-py3-none-any.whl (45.3 kB view details)

Uploaded Python 3

File details

Details for the file unitport-0.0.4.tar.gz.

File metadata

  • Download URL: unitport-0.0.4.tar.gz
  • Upload date:
  • Size: 42.7 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.7

File hashes

Hashes for unitport-0.0.4.tar.gz
Algorithm Hash digest
SHA256 97b3e91d157e5df2d8bef5649519553df337f386ae276bb369b739e5c40e5e8d
MD5 9c665ec990562e83f32aec6b90fa05a2
BLAKE2b-256 fd5b3f1f9ae20606c07ea2c94ea2a6b2f96c70a210ea9db7065a1e98c3e799fe

See more details on using hashes here.

File details

Details for the file unitport-0.0.4-py3-none-any.whl.

File metadata

  • Download URL: unitport-0.0.4-py3-none-any.whl
  • Upload date:
  • Size: 45.3 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.7

File hashes

Hashes for unitport-0.0.4-py3-none-any.whl
Algorithm Hash digest
SHA256 3f169890afe85ae262532359e537aa39a41474baad8dbe2eaf7ce43b622e681a
MD5 f2187167e4b813e79544d3d2244abbcb
BLAKE2b-256 7c974e84373de1e3fa61a04c9fe4d156aa9027dddbe2d96ef9aca739acfa88ab

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page