High-Risk Behavior Identification & Analysis Tool
Paste the following prompt into your AI chat to install this skill:
Please install @user_bb47e3e3/risk-analysis according to https://skillhub.cn/install/skillhub.md.
About this skill
Problem Solved
High-risk events involving older adults, chronic disease patients, or unmonitored users often rely on manual supervision or reviewing footage after the fact. This skill turns high-risk behavior review into a scriptable analysis flow for images, videos, local files, and network URLs, helping detect falls, prolonged stillness, abnormal behavior, and visual precursors of cardiac or cerebrovascular events while producing structured, traceable reports.
How It Works
The core capabilities cover fall, behavior, health, and all modes, using visual cues, pose estimation, and multimodal health-risk recognition to produce detections, alert levels, and recommendations. The typical workflow is: prepare an input source, run the analysis, retrieve the report link, and query historical reports. The common entrypoint is python -m scripts.risk_analysis, with parameters such as --input, --url, --list, --mode, --threshold, and --alert. Historical reports are read only from the cloud API and returned as a Markdown table; identity parameters are handled internally and do not need to be supplied by the user.
Boundaries
Supported inputs include mp4, avi, mov, jpg, png, rtsp, and http/https, with a maximum video size of 200MB. Outputs are assistive judgments and cannot replace professional security monitoring or medical diagnosis. Real-time stream alerts should still be followed by human confirmation and response.
Use Cases
- Community caregivers receive a camera screenshot or short video from a resident and need to judge whether an older adult has fallen or stayed still for too long, leaving a traceable conclusion.
- Security operations staff ingest local recordings or video URL inputs into the `risk_analysis` script to produce structured reports on falls, abnormal behavior, and health precursors.
- Chronic care personnel monitor a real-time stream URL for abnormal activity and review alert levels, risk types, and report links when high-risk behavior is detected.
- Operations administrators query historical risk analysis records for a period, using `--list` to pull paginated reports from the cloud API and verify them in a table.
Best For
- Caregivers for independent older adults: quickly judge whether screenshots or videos show a fall, prolonged stillness, and produce reviewable alert records.
- Security operations engineers: ingest local files, network URLs, or real-time streams into the script and output risk levels using `fall`, `behavior`, `health`, or `all` modes.
- Chronic disease managers: assist in identifying precursors of cardiac or cerebrovascular events, and combine report links and alert levels for manual review.
- Platform operations administrators: query paginated historical reports from the cloud API instead of extracting records from local memory or manual summaries.
Related Skills
Covers literature review, data analysis, visualization, drug discovery, paper writing, and grant writing workflows.
A retrieval assistant for Chinese laws and regulations, supporting semantic queries, validity checks, clause location, and regulation comparison with CLI-based output.
Run questionnaire-based tax risk scans that generate red/orange/yellow alerts with industry templates, policy references, remediation actions, and local trend tracking.
A contract review assistant for legal and business teams that extracts clauses, flags risks, compares versions, and generates opinion reports with a built-in legal knowledge base.