yolo-expert
PassAudited by ClawScan on May 7, 2026.
Overview
This appears to be a normal YOLO/Ultralytics reference skill, with disclosed examples for updating source code and running ML workflows that users should approve before executing.
This skill looks safe to use as a YOLO/Ultralytics reference. Before letting an agent run any examples, confirm the target files, datasets, output folders, and whether it may clone or update the Ultralytics repository.
Findings (2)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Answers or examples may depend on the latest upstream Ultralytics code, and running the setup writes a local repository checkout.
The skill documents cloning and updating an external GitHub repository without pinning a specific commit or version. This is purpose-aligned for source-code reference, but the reviewed source can change over time.
git clone --depth 1 https://github.com/ultralytics/ultralytics.git /root/ultralytics cd /root/ultralytics && git pull
Approve clone or pull commands before running them, and pin a trusted Ultralytics version or commit if reproducibility matters.
If these examples are run, they may create or replace local output files and use substantial CPU/GPU time.
The skill provides examples that can write files, overwrite existing experiment outputs, export model artifacts, and consume significant compute when executed. These operations are expected for YOLO workflows but should be user-directed.
results = model.train(... project="runs/detect", name="train", exist_ok=True, ...)
results[0].save_txt("labels/")
results[0].save_crop("crops/")
model.export(format="onnx", imgsz=640, half=True, opset=12)Run training, export, and save operations only on intended datasets and paths, and avoid overwrite options unless the user confirms.