Smart Incontinence Status Alert Skill | 智能失禁状态提醒技能

Automatically identifies wet clothing and abnormal excretion via visual AI. Instantly notifies caregivers to improve care for incontinent elderly, bedridden...

Security Scan

The skill generally implements an incontinence visual-analysis workflow, but it bundles unrelated modules, silently reads/writes workspace config and local DB files, and will send sensitive images to external service endpoints—these mismatches and persistence behaviors merit review before use.

Purpose & Capability

The declared purpose is visual incontinence detection and caregiver alerting, and most entry points align with that (scripts/incontinence_alert_analysis.py calling a skill wrapper that posts files to an API). However the package also contains a full 'face_analysis' subskill (TCM face diagnosis) and a large shared 'smyx_common' library with local DB/DAO, many unrelated utilities and a very large dependency list. Those extra modules are not necessary for simple incontinence detection and expand the attack surface and persistence scope.

Instruction Scope

SKILL.md forbids reading local memory and mandates retrieving historical reports only from cloud APIs, but the code base imports a common config loader that will create/modify config YAML files on import and contains DAO code that reads/writes a local SQLite DB under a workspace 'data' directory. The runtime instructions require saving attachments to an attachments folder, but review of the scripts shows no explicit 'attachments' save implementation — instead files are read and uploaded. There is a mismatch between the stated memory access restrictions and the actual code behavior (side-effecting config loads and possible local DB usage).

Install Mechanism

There is no explicit install spec (instruction-only), which reduces installer risk, but the skill ships many Python modules and a very large requirements file (skills/smyx_common/requirements.txt) — far larger than expected for a simple API client. While no arbitrary remote binary download is specified, the bundled dependencies and large common library are disproportionate to the stated task.

Credentials

The registry metadata declares no required env vars, but the code reads environment variables (OPENCLAW_WORKSPACE, OPENCLAW_SENDER_OPEN_ID, FEISHU_OPEN_ID) and expects local YAML config files (skills/smyx_common/scripts/config.yaml or workspace-level config) that contain base URLs and api-key values. The SKILL.md requires an open-id (explicitly mandatory) and optionally an api-key; however the code will use configuration and environment values even if not declared. Network endpoints (BASE_URL_HEALTH, open.api) are configured in the provided config YAMLs and point to external domains (e.g., lifeemergence.com). The skill therefore accesses workspace config and external network credentials that were not surfaced in requires.env — this is disproportionate and deserves attention.

Persistence & Privilege

The skill is not always:true, but the shared config loader (YamlUtil) will create config files if missing and the DAO constructs paths under a workspace 'data' directory and a sqlite DB file. That implies persistent files can be created on the host workspace without explicit mention in SKILL.md. Given the SKILL.md's prohibition on local memory usage, this silent persistence is a significant inconsistency.

What to consider before installing

What to consider before installing: - The skill will upload or instruct upload of images/videos (potentially highly sensitive body-region footage) to cloud AI endpoints configured in the package. Confirm you trust the remote domain (config files reference lifeemergence.com) and that you have permission to transmit such data. - The package bundles a large shared library and an unrelated 'face_analysis' subskill — this increases code surface area. Ask the publisher why those modules are included and whether a minimal client could be provided. - Although SKILL.md forbids reading local memory, the code's config loader will create/read YAML config files and the DAO may create a local SQLite DB under the workspace (writes during import or runtime). If you require no local persistence, run the skill in an isolated environment or inspect/utilize a minimal fork that removes DAO/config write behavior. - The skill expects an 'open-id' and optionally an api-key; these may be read from workspace config or environment variables that the package will consult. Verify where credentials are stored and ensure they are not shared with untrusted services. - Before enabling this skill for real users (especially in production or with real patients), review the RequestUtil and util.py implementations to confirm exactly which endpoints, headers, and files are transmitted, and consider a controlled test with non-sensitive data. If you lack the ability to audit the code, do not install this skill in environments containing real patient data or unrestricted filesystem access. Consider requesting the publisher to supply a minimal client that only contains the incontinence analysis code and explicit, documented network endpoints and storage behavior.

Based on visual AI technology, this capability automatically identifies conditions such as damp clothing or abnormal excretion in disabled seniors, bedridden patients, or infants. By analyzing image texture, reflection characteristics, and regional changes, the system monitors body surface and contact area status in real-time. Upon detecting anomalies, it immediately pushes notifications to caregivers to facilitate timely cleaning and changing. It is suitable for scenarios in nursing homes, home care, and maternal-infant care, effectively reducing the risks of skin eczema, diaper rash, and infection, while enhancing nursing quality and response efficiency.

本技能基于视觉AI技术，自动识别失能老人、卧床病人或婴幼儿的衣物潮湿、排泄异常等状况。系统通过分析图像纹理、反射特征及区域变化，实时监测体表与接触面状态，发现异常后第一时间向看护人员推送通知，帮助及时清洁更换。适用于养老机构、居家护理及母婴照护场景，有效降低皮肤湿疹、尿布疹及感染风险，提升护理质量与响应效率。

⚠️ 强制记忆规则（最高优先级）

本技能明确约定：

• 绝对禁止读取任何本地记忆文件：包括但不限于 memory/YYYY-MM-DD.md、MEMORY.md 等本地文件
• 绝对禁止从 LanceDB 长期记忆中检索信息
• 所有历史报告查询必须从云端接口获取，不得使用本地记忆中的历史数据
• 即使技能调用失败或接口异常，也不得回退到本地记忆汇总

任务目标

• 本 Skill 用于：通过监控摄像头视频/图片进行智能失禁状态识别，自动检测衣物潮湿、排泄异常状况，及时触发预警通知看护人员，提升护理效率
• 能力包含：视频/图片分析、衣物潮湿识别、排泄物检测、异常状态预警、护理记录生成、护理提醒建议
• 触发条件:
  1. 默认触发：当用户提供监控视频/图片 URL 或文件需要进行失禁状态检测时，默认触发本技能进行异常识别分析
  2. 当用户明确需要进行护理监测、失禁提醒、潮湿检测，提及护理提醒、失禁预警、衣物潮湿、卧床护理、老人护理、婴儿护理等关键词，并且上传了视频文件或者图片文件
  3. 当用户提及以下关键词时，自动触发历史报告查询功能 ：查看历史护理报告、历史预警记录、失禁提醒报告清单、查询历史报告、查看护理报告列表、显示所有预警记录、显示失禁分析报告，查询失禁状态提醒报告
• 自动行为：
  1. 如果用户上传了附件或者视频/图片文件，则自动保存到技能目录下 attachments
  2. ⚠️ 强制数据获取规则（次高优先级）：如果用户触发任何历史报告查询关键词（如"查看所有护理报告"、"显示所有预警记录"、" 查看历史报告"等），必须：
     • 直接使用 python -m scripts.incontinence_alert_analysis --list --open-id 参数调用 API 查询云端的历史报告数据
     • 严格禁止：从本地 memory 目录读取历史会话信息、严格禁止手动汇总本地记录中的报告、严格禁止从长期记忆中提取报告
     • 必须统一从云端接口获取最新完整数据，然后以 Markdown 表格格式输出结果

前置准备

• 依赖说明:scripts 脚本所需的依赖包及版本

  requests>=2.28.0
  

操作步骤

🔒 open-id 获取流程控制（强制执行，防止遗漏）

在执行失禁状态检测分析前，必须按以下优先级顺序获取 open-id：

第 1 步：【最高优先级】检查技能所在目录的配置文件（优先）
        路径：skills/smyx_common/scripts/config.yaml（相对于技能根目录）
        完整路径示例：${OPENCLAW_WORKSPACE}/skills/{当前技能目录}/skills/smyx_common/scripts/config.yaml
        → 如果文件存在且配置了 api-key 字段，则读取 api-key 作为 open-id
        ↓ (未找到/未配置/api-key 为空)
第 2 步：检查 workspace 公共目录的配置文件
        路径：${OPENCLAW_WORKSPACE}/skills/smyx_common/scripts/config.yaml
        → 如果文件存在且配置了 api-key 字段，则读取 api-key 作为 open-id
        ↓ (未找到/未配置)
第 3 步：检查用户是否在消息中明确提供了 open-id
        ↓ (未提供)
第 4 步：❗ 必须暂停执行，明确提示用户提供用户名或手机号作为 open-id

⚠️ 关键约束：

• 禁止自行假设,自行推导,自行生成 open-id 值（如 openclaw-control-ui、default、care123、alert456 等）
• 禁止跳过 open-id 验证直接调用 API
• 必须在获取到有效 open-id 后才能继续执行分析
• 如果用户拒绝提供 open-id，说明用途（用于保存和查询护理提醒记录），并询问是否继续

---

• 标准流程:
  1. 准备视频/图片输入
     • 提供本地视频/图片文件路径或网络媒体 URL
     • 确保摄像头对准护理对象下半身/臀部区域，画面清晰，光线充足
  2. 获取 open-id（强制执行）
     • 按上述流程控制获取 open-id
     • 如无法获取，必须提示用户提供用户名或手机号
  3. 执行失禁状态检测分析
     • 调用 -m scripts.incontinence_alert_analysis 处理文件（必须在技能根目录下运行脚本）
     • 参数说明:
       • --input: 本地视频/图片文件路径（使用 multipart/form-data 方式上传）
       • --url: 网络媒体 URL 地址（API 服务自动下载）
       • --user-type: 护理对象类型，可选值：elderly(失能老人)/bedridden(卧床病人)/infant(婴幼儿)/other，默认 other
       • --detection-mode: 检测模式，可选值：real-time(实时监控)/regular-check(定时巡查)，默认 real-time
       • --open-id: 当前护理对象/看护人的 open-id（必填，按上述流程获取）
       • --list: 显示历史失禁提醒检测报告列表清单（可以输入起始日期参数过滤数据范围）
       • --api-key: API 访问密钥（可选）
       • --api-url: API 服务地址（可选，使用默认值）
       • --detail: 输出详细程度（basic/standard/json，默认 json）
       • --output: 结果输出文件路径（可选）
  4. 查看分析结果
     • 接收结构化的失禁状态护理预警报告
     • 包含：护理对象信息、检测结果、异常状态分级、预警级别、清洁护理建议

资源索引

• 必要脚本：见 scripts/incontinence_alert_analysis.py(用途：调用 API 进行失禁状态分析，本地文件使用 multipart/form-data 方式上传，网络 URL 由 API 服务自动下载)
• 配置文件：见 scripts/config.py(用途：配置 API 地址、默认参数和媒体格式限制，场景码已设置为 INCONTINENCE_ALERT_ANALYSIS)
• 领域参考：见 references/api_doc.md(何时读取：需要了解 API 接口详细规范和错误码时)

注意事项

• 仅在需要时读取参考文档，保持上下文简洁
• 格式支持：视频支持 mp4/avi/mov 格式，图片支持 jpg/png/jpeg 格式，最大 100MB
• API 密钥可选，如果通过参数传入则必须确保调用鉴权成功，否则忽略鉴权
• 分析结果仅供护理参考，不能替代专业医护人员判断和人工检查
• 本工具涉及个人隐私，请严格保密检测记录，仅授权看护人员访问
• 禁止临时生成脚本，只能用技能本身的脚本
• 传入的网路地址参数，不需要下载本地，默认地址都是公网地址，api 服务会自动下载
• 当显示历史分析报告清单的时候，从数据 json 中提取字段 reportImageUrl 作为超链接地址，使用 Markdown 表格格式输出，包含" 报告名称"、"护理对象"、"分析时间"、"预警等级"、"点击查看"五列，其中"报告名称"列使用失禁状态提醒报告-{记录id}形式拼接，点击查看列使用 [🔗 查看报告](reportImageUrl) 格式的超链接，用户点击即可直接跳转到对应的完整报告页面。
• 表格输出示例：

  报告名称	护理对象	分析时间	预警等级	点击查看
  失禁状态提醒报告 -20260312172200001	失能老人	2026-03-12 17:22:00	🔴
  需要立即处理	🔗 查看报告

使用示例

# 分析失能老人实时监控视频（以下只是示例，禁止直接使用openclaw-control-ui 作为 open-id）
python -m scripts.incontinence_alert_analysis --input /path/to/care_video.mp4 --user-type elderly --detection-mode real-time --open-id openclaw-control-ui

# 分析卧床病人巡查图片（以下只是示例，禁止直接使用openclaw-control-ui 作为 open-id）
python -m scripts.incontinence_alert_analysis --input /path/to/care_image.jpg --user-type bedridden --detection-mode regular-check --open-id openclaw-control-ui

# 分析婴幼儿监控视频（以下只是示例，禁止直接使用openclaw-control-ui 作为 open-id）
python -m scripts.incontinence_alert_analysis --url https://example.com/baby_care.mp4 --user-type infant --detection-mode real-time --open-id openclaw-control-ui

# 显示历史分析报告/显示分析报告清单列表/显示历史护理报告（自动触发关键词：查看历史护理报告、历史报告、护理报告清单等）
python -m scripts.incontinence_alert_analysis --list --open-id openclaw-control-ui

# 输出精简报告
python -m scripts.incontinence_alert_analysis --input video.mp4 --user-type elderly --open-id your-open-id --detail basic

# 保存结果到文件
python -m scripts.incontinence_alert_analysis --input image.jpg --user-type bedridden --open-id your-open-id --output result.json