cdf-anomalygen

77 / 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

Use when running the CDF AnomalyGen pipeline over a defect dataset — mask placement, fine-tuning, synthetic defect-image generation (SDG), evaluation, quality refinement, and pseudo-labeling — even when the user names only one stage.

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/cdf-anomalygen/SKILL.md, line 58scripts/preflight_env_ckpt.sh
skills/cdf-anomalygen/SKILL.md, line 76[ -f ~/.anomalygen/endpoints.yaml ] && echo "amg model: ~/.anomalygen/endpoints.yaml present" \
skills/cdf-anomalygen/SKILL.md, line 166source scripts/skill_utility/set_pipeline_vars.sh --name <name> --dataset_dir <dataset_dir> \

Show 15 more

skills/cdf-anomalygen/SKILL.md, line 188set -o pipefail
skills/cdf-anomalygen/SKILL.md, line 202RUN_DIR=<existing run dir> # only if it is not the ${RUN_DIR} already set above
skills/cdf-anomalygen/SKILL.md, line 225python -m anomalygen.scripts.datasets.prepare_phone_screen_defect --output_dir "${DATASET_DIR}" --zip <roboflow.zip>
skills/cdf-anomalygen/SKILL.md, line 249grep -l spatial_dependency "${DEFECT_SPEC}" >/dev/null 2>&1 && echo "NOTE: spatial_dependency present but mask_backend=amg ignores it; use mask_backend=amp to
skills/cdf-anomalygen/SKILL.md, line 263AMG_MODE=validation; AMG_DIR="${VAL_DIR}"; ALLOC=validation # names the log, the out-dir, the split
skills/cdf-anomalygen/SKILL.md, line 318scripts/agentic_mask_generation_serving/serverctl.sh health # idempotent: probe first, start only if down
skills/cdf-anomalygen/SKILL.md, line 354AMP_MODE=validation; AMP_DIR="${VAL_DIR}"; SEED=42 # or: AMP_MODE=inference AMP_DIR="${GEN_DIR}" SEED=43
skills/cdf-anomalygen/SKILL.md, line 385TEMPLATE=ag_config/exp_texture_ft_phone_screen.yaml # the worked example doubles as the template
skills/cdf-anomalygen/SKILL.md, line 423scripts/skill_utility/train_control.sh start --name "${NAME}" --recipe "${RECIPE}" \
skills/cdf-anomalygen/SKILL.md, line 440scripts/skill_utility/train_control.sh wait --name "${NAME}" # 0 = finished · 1 = failed · 2 = still running, call again
skills/cdf-anomalygen/SKILL.md, line 454CKPT="${RUN_DIR}/checkpoints/model/$(cat "${RUN_DIR}/checkpoints/best_checkpoint.txt")" # iter_<best>.pt
skills/cdf-anomalygen/SKILL.md, line 469torchrun --nproc_per_node="${NUM_GPUS}" "${SCRIPTS}/generate.py" \
skills/cdf-anomalygen/SKILL.md, line 494python "${SCRIPTS}/evaluate.py" \
skills/cdf-anomalygen/SKILL.md, line 518python "${SCRIPTS}/quality_refine.py" run \
skills/cdf-anomalygen/SKILL.md, line 565python "${SCRIPTS}/pseudo_label.py" \

Changes files outside its own folder

The author does not mention this.

Found in 1 place.

skills/cdf-anomalygen/SKILL.md, line 98`mkdir -p ~/.anomalygen && cp scripts/agentic_mask_generation_serving/endpoints.example.yaml ~/.anomalygen/endpoints.yaml` wires it: the template's gemma entry

Can call paid AI services billed to you

The author does not mention this.

Found in 3 places.

skills/cdf-anomalygen/references/agentic-mask-generation.md, line 48(`https://api.anthropic.com/v1`, the one base constant in the code) serves exactly one protocol —
skills/cdf-anomalygen/references/agentic-mask-generation.md, line 66(`https://.../v1`, Gemini's `https://generativelanguage.googleapis.com/v1beta/openai`, ...) — put
skills/cdf-anomalygen/references/agentic-mask-generation.md, line 73gateways translate it per backend — so the shipped cloud model works from `https://api.openai.com/v1`

Install

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

if [ -e ~/.agents/skills/'cdf-anomalygen' ] || [ -L ~/.agents/skills/'cdf-anomalygen' ]; then echo 'A skill folder named cdf-anomalygen 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 "$d" checkout --quiet '2630133f22e357f2005886567a39eddcc1b520e2' -- 'skills/cdf-anomalygen' && mkdir -p ~/.agents/skills && cp -R "$d"/'skills/cdf-anomalygen' ~/.agents/skills/'cdf-anomalygen' && 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/'cdf-anomalygen' ~/.claude/skills/'cdf-anomalygen' (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`	77
2026-11-08	`fce3cca`	77
2026-11-07	`fce3cca`	77