tao-run-automl

64 / 100 · Review before installing

Check before installing. Some of what it can do is expected: a coding skill usually runs build and test commands.

By NVIDIA/skills (GitHub) · coding · 3,933 stars · scanned at commit 2630133

Run container-backed AutoML / hyperparameter optimization (HPO) for NVIDIA TAO networks using AutoMLRunner. Handles algorithm

Works with any agent that supports SKILL.md skills.

What this skill can do

Runs commands on your computer

This is disclosed by the author in the skill's SKILL.md or README. Expected for a coding skill.

Found in 18 places.

skills/tao-run-automl/SKILL.md, line 83python -c "import tao_automl; from tao_automl.runner import AutoMLRunner; print('OK')"
skills/tao-run-automl/SKILL.md, line 95SB="${TAO_SKILL_BANK_PATH:-~/tao-skill-bank}"
skills/tao-run-automl/cards/00-preflight.md, line 5bash -c 'grep -h "ok$\|FAIL$" $RD/progress.log 2>/dev/null | tail -3; echo "---GATE-END---"'

Show 15 more

skills/tao-run-automl/cards/00-preflight.md, line 21bash -c 'docker image inspect $TRAIN_IMG --format "image OK: {{.Id}}" | head -1 && nvidia-smi --query-gpu=name,memory.total --format=csv,noheader'
skills/tao-run-automl/cards/00-preflight.md, line 26bash -c 'grep -m1 "automl_enabled" $SB/skills/models/tao-train-visual-changenet/references/skill_info.yaml; $VENV/bin/python3 -c "import json;json.load(open(\"$
skills/tao-run-automl/cards/00-preflight.md, line 32bash -c 'mkdir -p $RD/{runner,review,baseline,automl_workspace}; touch $RD/commands.log; [ -f $RD/state.json ] || echo "{}" > $RD/state.json; echo "preflight ok
skills/tao-run-automl/cards/10-baseline-eval.md, line 10bash -c 'grep -h "baseline_eval ok" $RD/progress.log 2>/dev/null | tail -1; [ -f $RD/baseline/BASELINE_JOB_DONE ] && echo MARKER_DONE; [ -f $RD/baseline/job.pid
skills/tao-run-automl/cards/10-baseline-eval.md, line 22bash -c 'cp $WS/specs/baseline_spec.yaml $RD/baseline/baseline_lr0_spec.yaml && sed -i -E -e "0,/^results_dir:/s|^results_dir:.*|results_dir: /results|" -e "s|^
skills/tao-run-automl/cards/10-baseline-eval.md, line 28bash -c 'printf "#!/bin/bash\ndocker run --rm --gpus all --shm-size=8g -v $WS:/data/workspace -v $RD/baseline:/results -v $WS/kpi/images:/data/datasets/NV_PCB_S
skills/tao-run-automl/cards/10-baseline-eval.md, line 34bash -c 'set -e; tail -2 $RD/baseline/train.log | grep -q "Execution status: PASS" || { echo "baseline_eval FAIL" >> $RD/progress.log; exit 0; }; $VENV/bin/pyth
skills/tao-run-automl/cards/20-launch-recs.md, line 5bash -c 'grep -hE "recs_launched ok|baseline_eval ok" $RD/progress.log 2>/dev/null | sort -u; echo ---GATE-END---'
skills/tao-run-automl/cards/20-launch-recs.md, line 21bash -c 'BL=$($VENV/bin/python3 -c "import json;print(json.load(open(\"$RD/state.json\")).get(\"baseline_val_loss\",\"MISSING\"))"); printf "# AutoML Launch Rev
skills/tao-run-automl/cards/20-launch-recs.md, line 26bash -c 'cat > $RD/runner/driver.py <<PYEOF
skills/tao-run-automl/cards/20-launch-recs.md, line 106bash -c 'cd $RD/runner && nohup $VENV/bin/python3 $RD/runner/driver.py > $RD/runner/driver.log 2>&1 & echo "runner PID $!"; sleep 20; tail -8 $RD/runner/driver.
skills/tao-run-automl/cards/30-interpret.md, line 5bash -c 'grep -h "done ok" $RD/progress.log 2>/dev/null | tail -1; [ -f $RD/automl_result.json ] && echo RESULT_EXISTS; pgrep -f "$RD/runner/driver.py" >/dev/nu
skills/tao-run-automl/cards/30-interpret.md, line 13| none (runner dead, no result, no FAIL) | runner was killed mid-flight: relaunch with resume — `bash -c 'cd $RD/runner && AUTOML_RESUME=1 nohup $VENV/bin/pytho
skills/tao-run-automl/cards/30-interpret.md, line 54bash -c 'echo "done ok" >> $RD/progress.log && touch $RD/AUTOML_DONE.marker && tail -3 $RD/progress.log'
skills/tao-run-automl/tests/test_resolve_automl_session.py, line 99result = subprocess.run(

Downloads and installs other software

The author does not mention this.

Found in 14 places.

skills/tao-run-automl/SKILL.md, line 96pip install "$($SB/scripts/resolve_versions_key.py wheels.tao_automl_<platform>)"
skills/tao-run-automl/eval.config, line 25"description": "NVIDIA Artifactory PyPI user. versions.yaml pins internal RC wheels (nvidia-tao-automl 7.1.0rc33, nvidia-tao-sdk 7.1.0rc42) that are NOT publish
skills/tao-run-automl/references/automl-advanced-monitoring.md, line 42512. **`openai` package not installed.** The `llm`, `hybrid`, and `autoresearch` algorithms require the `openai` Python package. Install with `pip install openai

Show 11 more

skills/tao-run-automl/references/automl-preflight-concepts.md, line 46echo " pip install \"$($SB/scripts/resolve_versions_key.py wheels.tao_automl_slurm)\" # on-prem SLURM cluster"
skills/tao-run-automl/references/automl-preflight-concepts.md, line 47echo " pip install \"$($SB/scripts/resolve_versions_key.py wheels.tao_automl_kubernetes)\" # K8s (EKS / GKE / on-prem)"
skills/tao-run-automl/references/automl-preflight-concepts.md, line 48echo " pip install \"$($SB/scripts/resolve_versions_key.py wheels.tao_automl_docker)\" # local Docker daemon"
skills/tao-run-automl/references/automl-preflight-concepts.md, line 49echo " pip install \"$($SB/scripts/resolve_versions_key.py wheels.tao_automl_brev)\" # Brev GPU instances"
skills/tao-run-automl/references/automl-preflight-concepts.md, line 50echo " pip install \"$($SB/scripts/resolve_versions_key.py wheels.tao_automl_all)\" # all platforms"
skills/tao-run-automl/references/automl-preflight-concepts.md, line 56(For local development against a checkout: `pip install -e '~/tao-run-automl[brev]'` from the cloned repo.)
skills/tao-run-automl/references/automl-preflight-concepts.md, line 125pip install "$($SB/scripts/resolve_versions_key.py wheels.tao_automl_brev)" # or _slurm, _kubernetes, _docker, _all
skills/tao-run-automl/references/automl-preflight-concepts.md, line 127pip install "$($SB/scripts/resolve_versions_key.py wheels.tao_automl_brev | sed 's/]/,llm]/')"
skills/tao-run-automl/references/automl-preflight-concepts.md, line 129For local development against a checkout: `pip install -e '~/tao-run-automl[brev]'`.
skills/tao-run-automl/references/automl-runner-configuration.md, line 28- `pip install "$(${TAO_SKILL_BANK_PATH:?}/scripts/resolve_versions_key.py wheels.tao_automl_brev | sed 's/]/,wandb]/')"`
skills/tao-run-automl/references/automl-runner-configuration.md, line 334pip install wandb

Reads your saved credentials, such as SSH keys or .env files

This is disclosed by the author in the skill's SKILL.md or README.

Found in 3 places.

skills/tao-run-automl/config/skillspector-baseline.yaml, line 12reason: "Reviewed FP (#67): documented infrastructure path reference (e.g. ~/.docker/config.json, ~/.ssh keys) appearing in documentation or example config; the
skills/tao-run-automl/eval.config, line 5"prompt": "Run TAO AutoML to tune image-classification training on my dataset.\n\n## Mandatory completion\nThis eval grades end-to-end completion. Do NOT stop a
skills/tao-run-automl/eval.config, line 21"description": "S3 secret key paired with ACCESS_KEY. The runner writes it into ~/.aws/credentials but does NOT export AWS_SECRET_ACCESS_KEY; the prompt exports

Changes files outside its own folder

This is disclosed by the author in the skill's SKILL.md or README.

Found in 3 places.

skills/tao-run-automl/eval.config, line 5"prompt": "Run TAO AutoML to tune image-classification training on my dataset.\n\n## Mandatory completion\nThis eval grades end-to-end completion. Do NOT stop a
skills/tao-run-automl/references/automl-preflight-concepts.md, line 56(For local development against a checkout: `pip install -e '~/tao-run-automl[brev]'` from the cloned repo.)
skills/tao-run-automl/references/automl-preflight-concepts.md, line 129For local development against a checkout: `pip install -e '~/tao-run-automl[brev]'`.

Connects to the internet

This is disclosed by the author in the skill's SKILL.md or README.

Found in 1 place.

skills/tao-run-automl/references/automl-advanced-monitoring.md, line 396- **LLM endpoint unreachable** (llm/hybrid/autoresearch only) — the brain falls back to random sampling. Check `AUTOML_LLM_ENDPOINT` and `AUTOML_LLM_API_KEY`. V

Safer alternatives

Low-risk skills in the same category (coding) that look related. Check that one does what you need.

Install

Paste this into your terminal. It downloads exactly the version we scanned, not whatever is newest.

if [ -e ~/.agents/skills/'tao-run-automl' ] || [ -L ~/.agents/skills/'tao-run-automl' ]; then echo 'A skill folder named tao-run-automl already exists — not replacing it' >&2; false; else d=$(mktemp -d) && git clone --quiet --filter=blob:none --no-checkout 'https://github.com/NVIDIA/skills' "$d" && git -C "$d" checkout --quiet '2630133f22e357f2005886567a39eddcc1b520e2' -- 'skills/tao-run-automl' && mkdir -p ~/.agents/skills && cp -R "$d"/'skills/tao-run-automl' ~/.agents/skills/'tao-run-automl' && rm -rf "$d"; fi

Codex reads ~/.agents/skills directly. For Claude Code, Cursor, Copilot, Gemini CLI, Antigravity or OpenCode, also link it. For Claude Code: ln -s ~/.agents/skills/'tao-run-automl' ~/.claude/skills/'tao-run-automl' (other agents: ~/.cursor/skills, ~/.copilot/skills, ~/.gemini/skills, ~/.gemini/config/skills, ~/.config/opencode/skills). Or install through the finder skill, which does both.

Badge for your README

The badge for NVIDIA/skills shows the score of its lowest-scoring skill.

SkillTrust badge for NVIDIA/skills

[![SkillTrust](https://zalavijaysinh.github.io/skilltrust/badges/NVIDIA/skills.svg)](https://zalavijaysinh.github.io/skilltrust/repos/NVIDIA/skills/)

Score history

Date	Version	Score
2026-11-11	`2630133`	64
2026-11-08	`fce3cca`	64
2026-11-07	`fce3cca`	64