Infant Nighttime Blanket Kick Monitoring

Problem

Infants can kick off blankets and stay uncovered during sleep, a risk that is easy to miss and may lead to chilling. Manual monitoring or after-the-fact video review does not clearly answer whether the baby kicked the blanket, how long exposure lasted, or whether a reminder is needed. This skill focuses on infant crib monitoring video and narrows the task to blanket-kicking and covering-risk detection, producing readable monitoring conclusions and report links.

How It Works

• Inputs include local video files, network URLs, images, or related attachments; common formats are mp4, avi, and mov, with a documented maximum of 10MB.
• Core analysis is executed by the scripts.infant_blanket_kick_monitoring_analysis script; local files can be passed with --input, while URLs can be sent with --url for cloud-side download.
• The monitoring chain covers infant pose recognition, blanket coverage detection, body-exposure judgment, blanket-kicking behavior recognition, and alert triggering, then returns a structured report, risk notes, suggestions, and report links.
• Historical report queries should be read from the cloud API using --list, not from local memory or manual summaries; returned JSON should be rendered as a Markdown table with report fields converted into hyperlinks.
• User identity is associated internally by the system; replies should not ask for, expose, or list usernames, phone numbers, or internal identity values.

Boundaries

• The camera should remain fixed and cover the whole crib area, with night vision or infrared capable of making the baby's outline clear after lights are off.
• Keep the crib free from excessive clutter so the baby's body and blanket area are fully visible.
• Monitoring output is only an auxiliary reminder and does not replace parental supervision or infant safety monitoring.
• Only the skill's built-in scripts should be used, avoiding ad hoc scripts or exposing internal parameters to users.