What problem does it solve? Installing the BrainX ecosystem correctly is error-prone: users must pick a coherent release whose component versions are compatible, choose the right device extra (cpu, cuda12, cuda13, tpu), and avoid mutating the wrong Python environment. This Skill enforces a confirmation-gated workflow that inspects the environment, selects one indivisible BrainX release, and verifies the result before reporting. ## Core Features & Use Cases - Guided installation and upgrade: Performs read-only environment inspection, requires explicit device-target confirmation, resolves the latest Python-compatible BrainX release, and installs only the confirmed meta-package release. - Compatibility-matrix release matching: When BrainX components already exist or a pinned version is requested, uses the bundled version matrix to select one coherent release instead of mixing component versions. - Safe uninstall and cleanup: Removes BrainX ecosystem packages (and optionally JAX distributions) from one confirmed virtual environment while preserving shared dependencies, with reverse-dependency checks and pip-check verification. - Use Case: A researcher with an NVIDIA GPU asks to set up BrainX. The Skill detects CUDA evidence, asks the user to confirm cuda12, presents the full installation specification, and only after approval creates the venv, installs BrainX[cuda12], and verifies JAX sees the GPU. ## Quick Start Ask the agent to install BrainX for your device target, for example: install BrainX with CUDA 12 support in a new virtual environment and verify it works.
