Run a Python training/eval script directly in an existing local virtualenv — no docker, no container. Implements the four-verb consumer contract (submit/status/logs/cancel) over a vendored process-lifecycle runner with durable on-disk state, PID-reuse-safe identity, and process-group cleanup.
Install
npx skills add NVIDIA/skills --skill tao-run-on-virtualenvnpx skills add NVIDIA/skills --skill tao-run-on-virtualenv -a claude-code -gnpx skills add NVIDIA/skills --skill tao-run-on-virtualenv -a codex -ggemini skills install https://github.com/NVIDIA/skills.git --path skills/tao-run-on-virtualenvgh skill install NVIDIA/skills tao-run-on-virtualenvgit clone --depth 1 https://github.com/NVIDIA/skills /tmp/NVIDIA__skills
mkdir -p ~/.claude/skills ~/.agents/skills
cp -R /tmp/NVIDIA__skills/skills/tao-run-on-virtualenv ~/.claude/skills/tao-run-on-virtualenv # Claude Code
cp -R /tmp/NVIDIA__skills/skills/tao-run-on-virtualenv ~/.agents/skills/tao-run-on-virtualenv # Codex, Cursor, Gemini CLI, Copilot, OpenCode