Use when uploading images or videos, labeling, organizing datasets, creating Roboflow projects (detection/segmentation/keypoint/classification), tags, splits, versions, or RoboQL search.
93%
Does it follow best practices?
Run evals on this skill
Adds up to 20 points to the overall score
Low
Low-risk findings worth noting
The skill exposes the agent to untrusted, user-generated content from public third-party sources, creating a risk of indirect prompt injection. This includes browsing arbitrary URLs, reading social media posts or forum comments, and analyzing content from unknown websites.
The required upload and search workflow ingests the user's own selected files, tenant data, and provider-generated status, but the optional "Universe fork" method (line 41) lets the agent pull in a community-published public dataset, whose images, annotations, and class names are outsider-authored and selected by the agent rather than forced on it.
roboflow/computer-vision-skills
c4263a7
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.