Health Feeling Logging Skill

Interactively record health feelings, converting natural language into structured JSON and saving.

核心流程

用户输入 → 解析信息 → 检查完整性 → [信息不足] 询问补充 → 生成JSON → 保存
                                          ↓
                                     [信息完整]

步骤 1: 解析用户输入

从用户输入中提取所有可用信息，映射到以下字段：

关键词映射

Body Part (bodyPart)

Pain Level (painLevel + severity)

Pain Type (painType)

步骤 2: 检查信息完整性

Must Have (ask if missing):

• symptom - Symptom description

Recommended (ask if missing):

• bodyPart - Body part (skip if can be inferred)
• painLevel / severity - Pain level (if pain is mentioned but no level given)

Optional (do not ask):

• painType - Pain type
• duration - Duration
• triggers - Triggers
• reliefFactors - Relief factors
• accompanyingSymptoms - Accompanying symptoms
• mood - Mood
• energyLevel - Energy level
• notes - Notes
• tags - Tags

步骤 3: 交互式询问（信息不足时）

采用面诊式对话风格：友好、自然、一次问一个问题。

对话原则

• 像医生问诊一样，用日常语言交流
• 每次只问 1-2 个相关问题
• 根据用户回答自然引导下一问
• 避免机械的选项罗列

询问场景示例

Scenario A: Missing Symptom Description

OK, can you tell me in detail where you're feeling uncomfortable?

Scenario B: Missing Body Part

Is it on the left, right, upper, or lower side of your abdomen?

Scenario C: Pain Mentioned But Level Missing

How severe is the pain? If we rate it on a scale of 1 to 10,
where 1 is almost no feeling and 10 is unbearable pain,
what would you rate it?

#### 场景 D: 多个缺失信息（逐步询问）

第1问: 哪里不舒服？ 用户: 肚子

第2问: 肚子的哪个位置？左边、右边、上面还是下面？ 用户: 左边

第3问: 疼得厉害吗？1-10分的话大概几分？ 用户: 3分吧

### Pain Level Reference (internal use, do not show to user)

| Level | Score | Description |
|-------|-------|-------------|
| Almost no sensation | 1 | Barely noticeable |
| Mild discomfort | 2 | Uncomfortable but doesn't affect activity |
| Mild pain | 3 | Slight pain, can be ignored |
| Moderate pain | 4 | Noticeable discomfort, occasionally noticed |
| Significant pain | 5 | Pain is obvious, affects attention |
| Heavy pain | 6 | Pain is significant, requires endurance |
| Severe pain | 7 | Pain is severe, affects daily activities |
| Intense pain | 8 | Intense pain, hard to concentrate |
| Extreme pain | 9 | Extreme pain, can barely do anything |
| Unbearable | 10 | Unbearable pain, needs immediate treatment |

## 步骤 4: 生成 JSON

信息完整后，生成符合 Schema 的 JSON：

```json
{
  "timestamp": "2026-02-03T10:30:00Z",
  "bodyPart": "left_abdomen",
  "symptom": "abdominal discomfort",
  "painLevel": "1",
  "severity": 1,
  "painType": "vague",
  "tags": ["pain", "abdomen"]
}

完整 Schema 定义参见 schema.json。

步骤 5: 保存数据

1. 读取 data/health-feeling-logs.json
2. 将新记录追加到数组末尾
3. 写回文件

执行指令

1. 解析用户输入，提取所有可用信息
2. 检查必填字段（symptom）和建议字段（bodyPart, painLevel）
3. 如有缺失，使用 AskUserQuestion 询问用户
4. 生成符合 schema.json 的 JSON
5. 追加保存到 data/health-feeling-logs.json
6. 向用户确认保存成功，并显示摘要

示例交互

完整输入（无需询问）

用户: 肚子不舒服，左腹，轻微疼痛
→ 直接解析并保存

不完整输入（需要询问）

用户: 肚子疼
AI: 哪个部位疼？（左腹、右腹、上腹、下腹？）
用户: 左腹
AI: 疼痛程度如何？（1-10级）
用户: 2级
→ 解析并保存

更多示例参见 examples.md。