AI Decision System
Replayable, offline-verifiable records for AI-assisted go/no-go decisions: experiment promotions, typed model answers and human-approved actions.
A real Linux workstation on an ARM64 phone — no PC, no root. Measured Zink/Turnip GPU route on Adreno, VirGL fallback. One command, auditable Bash.
Translation pending
Developer toolsShell · PythonGitHub @soobujmiahOnlineOwner unconfirmed
Have you tried it? One tap is enough. No sign-in needed.
Owners of public repositories under a personal GitHub account can sign in to verify ownership and manage the introduction themselves. No administrator approval is needed.
Sign in, generate a verification file, and commit it to this repository’s default branch within 30 minutes. Files on other branches or forks do not count. Then check the file below to manage the project without administrator approval. Keep the file to retain your access.
If needed, you will go to sign-in. After signing in, return here and press your chosen verification button again.
Run /nomorevibe verify in your project folder to verify ownership and manage the introduction yourself.
Verify in one minuteAsk about other verification options
A GPU-accelerated Linux desktop on your Android phone. Termux + PRoot Debian + Xfce4, with Zink/Turnip Vulkan acceleration. No root.
Taken from the repository README
Loading the full README. The saved excerpt is shown below.
Machine-generated operational state: see .repo/STATUS.md (generated by tools/repo_knowledge). For cross-project SKB registry state, see ternux.json in the parent SKB repository (soobujmiah/skb).
ternux
A real Linux workstation on an ARM64 phone — no PC, no root, verified graphics paths
ternux turns an ARM64 Android phone into a Debian + Xfce4 Linux workstation through Termux and PRoot. It provides Termux:X11 display, PulseAudio, a permanent ternux control CLI, and a graphics route using Zink → Turnip on supported Adreno devices or VirGL as a fallback.Read the full README ↗Machine-generated operational state: see .repo/STATUS.md (generated by tools/repo_knowledge). For cross-project SKB registry state, see ternux.json in the parent SKB repository (soobujmiah/skb).
Website · Deep documentation · বাংলা · Source
Ternux sets up Debian and Xfce4 in Termux using PRoot. Termux:X11 displays the desktop and PulseAudio connects sound. Graphics use Zink/Turnip on supported Adreno devices or VirGL as a compatibility route. Results vary by device.
curl -fsSL https://raw.githubusercontent.com/soobujmiah/ternux/main/install.sh | bash
If curl is unavailable, use:
wget -qO- https://raw.githubusercontent.com/soobujmiah/ternux/main/install.sh | bash
The base install needs about 3–4 GB of storage. Keep extra temporary space for downloads and package caches.
In Termux, start the desktop:
source ~/.bashrc
x
Switch to the Termux:X11 app to see Xfce. Common commands, all run in Termux:
ternux status # show session and install state
ternux verify # check the installation and graphics route
ternux doctor # diagnose common problems
ternux stop # stop the desktop session
The x shortcut starts the desktop; xgo opens Termux:X11 and starts it.
Inside Debian, use db from Termux to open a shell. Shared Android storage is
available at /sdcard inside Debian.
For Android app development on this same device, install ADT inside the Ternux Debian guest. The build-to-device workflow is validated on the shared Termux + PRoot host; use from inside the Ternux Xfce session is not formally tested yet. See the ADT setup steps and ADT + Ternux notes.
For Blender work, coding agents can use Ternux's local GPU MCP to check the render route and run scripts or scenes with the measured EEVEE setup. See the MCP guide and render evidence.
With a terminal attached, the one-command installer suggests optional workloads and lets you choose by number; Enter keeps the base desktop, while --yes skips prompts and installs the base profile. The website provides the same short install and usage path, explains the numbered
choices and summarizes practical device workloads; deeper technical procedures
remain in the repository guides. Current experimental CPU/GPU/NPU release gates
are tracked in docs/ACCELERATOR-RELEASE-PLAN.md.