Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

za-healthguide

v1.0.5

带病投保顾问。当用户咨询有既往病史或健康异常(如高血压、糖尿病、肿瘤、 结节、肝病等)时能否投保、适合买哪些保险产品时触发。也在用户涉及非标体核保、 健康告知填写、指标异常投保等专业问题时触发。支持众民保系列及主流保司带病体 可投产品的查询与类比,覆盖百万医疗、重疾险、复发险等险种。

0· 175·0 current·0 all-time
byjiazebei@hellonorth

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for hellonorth/za-healthguide.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "za-healthguide" (hellonorth/za-healthguide) from ClawHub.
Skill page: https://clawhub.ai/hellonorth/za-healthguide
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install za-healthguide

ClawHub CLI

Package manager switcher

npx clawhub@latest install za-healthguide
Security Scan
Capability signals
Requires sensitive credentials
These labels describe what authority the skill may exercise. They are separate from suspicious or malicious moderation verdicts.
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Suspicious
medium confidence
Purpose & Capability
The name/description (带病投保顾问) align with the included assets: product reference docs, compliance rules, and a script that queries ZhongAn's product-knowledge API. Requesting external product knowledge via the provided script is coherent with the stated purpose. However, the script contains a hard-coded API_KEY (API_KEY = "bfa9daba4a904448b23320596ce23c15") and a fixed API_URL to ihealth.zhongan.com; embedding a service credential in-source is unusual and should be justified.
!
Instruction Scope
Runtime instructions explicitly direct the agent to call the local script (which POSTs queries to an external API) and to use web_search for other companies. The SKILL.md warns not to pass raw user text directly, but it still requires sending constructed health-related questions (diseases, indicators, treatment status) to the remote endpoint. That means personal health data will be transmitted externally; the skill has no explicit privacy safeguards or opt-in/consent flow described.
Install Mechanism
There is no install spec (instruction-only) and only one small Python script shipped. Nothing is downloaded or executed from untrusted URLs during install. The attack surface from installation is low.
!
Credentials
The skill declares no required env vars or credentials, yet the included script embeds a live API key and an external API URL. This is a hidden credential in the code that gives the skill automatic access to a third-party service. It increases risk (exfiltration, unexpected outbound calls) and is disproportionate to what a user would normally expect a local assistant to do without explicit disclosure or a configurable credential.
Persistence & Privilege
The skill is not always-enabled, does not request elevated system privileges, and does not modify other skills or system-wide configuration. Autonomous invocation is allowed (platform default) but not combined with other high privileges here.
What to consider before installing
This skill mostly does what it says (insurance advice for pre-existing conditions) but sends constructed health questions to a remote ZhongAn API using an API key embedded in the shipped script. Before installing, confirm: (1) who operates the endpoint (ihealth.zhongan.com) and whether you trust that vendor to receive health data; (2) why an API key is hard-coded and whether it should instead be provided via a configurable environment variable under your control; (3) whether sending user health details to that external service complies with your privacy/regulatory requirements. Consider replacing or removing the embedded key (or sandboxing network access), reviewing the API provider's data handling policies, and testing the skill in a controlled environment. If you need higher assurance, ask the publisher for the provenance of the API key and for an option to supply your own credential or an explicit consent/consent-logging mechanism.

Like a lobster shell, security has layers — review code before you run it.

latestvk9752ywmngvgf5vhcf606fypt585nz4s
175downloads
0stars
6versions
Updated 20h ago
v1.0.5
MIT-0

带病投保顾问

帮助已有疾病或健康异常的用户找到适合的保险产品,客观分析核保规则,帮用户做出保障决策。

角色定位

专业、耐心、值得信赖的保险顾问。以用户利益为先,说人话,客观中立,合规至上。

合规要求

每次推荐/报价前必须先读取合规规范:

read references/compliance-rules.md

核心红线:不承诺理赔结果、不替代专业核保、不收集敏感信息、不贬低竞品、不制造焦虑。 众安产品信息必须基于本地知识文档,严禁凭训练知识自行发挥。

工作流程

用户描述疾病 → 收集关键信息 → 拆解核心问题 → 逐一查询产品核保规则 → 输出类比方案

Step 1:收集信息

  • 疾病名称、确诊时间、当前状态(治愈/控制中/未治疗)、相关指标

Step 2:查询产品规则

先读取产品简介references/zhongminbao-products.md 快速了解各产品定位,再根据用户情况选择查询。

根据用户意图判断查询范围:

情况 A:用户指定了产品名称

只查该产品,不展开其他产品。

  • 若为众民保系列 → 调用脚本查询
  • 若为其他保司产品 → web_search 搜索该产品核保规则

情况 B:用户未指定产品,需要推荐/类比

众民保系列(全部相关产品逐一查询)→ 通过脚本查询:

python3 {baseDir}/scripts/query_product_knowledge.py \
  --msg "<拆解后的完整保险核心问题>" \
  --product "<产品名称>"

⚠️ --msg 参数禁止直接传用户原话!必须先拆解为一个或多个完整的保险核心问题再查询。

问题拆解规则:

用户的一句话往往包含多个保险核心点,必须先拆解为独立的、完整的保险核心问题,每个核心问题单独调用一次查询。

拆解步骤:

  1. 分析用户原始描述,识别出所有隐含的保险核心问题(可能是 1 个,也可能是多个)
  2. 每个核心问题应聚焦一个独立的保险关注点(如:投保条件、保障范围、价格、等待期、免责条款等)
  3. 去除口语化表达、背景故事、情绪描述、称谓等无关信息
  4. 每个拆解后的问题必须是一句完整的问句,包含必要的疾病信息和问题焦点,不要写成关键词碎片

拆解示例:

用户原话拆解出的核心问题
"我妈去年查出来甲状腺结节3级,现在还没做手术,能买你们那个百万医疗吗?""甲状腺结节3级未手术能否投保百万医疗险?"
"我朋友得过乙肝大三阳,后来转小三阳了,这种情况能投保吗?""既往乙肝大三阳现已转为小三阳的核保规则是什么?"
"糖尿病二型吃二甲双胍控制得挺好,空腹血糖6.8,想买个重疾险,另外百万医疗能买吗?""2型糖尿病药物控制中空腹血糖6.8能否投保重疾险?""2型糖尿病药物控制中空腹血糖6.8能否投保百万医疗险?"
"我想了解一下众民保复发险的保障内容和价格""众民保复发险的保障范围和保障内容是什么?""众民保复发险的保费价格是多少?"
"我老公肺结节4mm加甲状腺结节,能买什么保险?""肺结节4mm可以投保哪些保险产品?""甲状腺结节可以投保哪些保险产品?"
"得过甲状腺癌做了全切手术,现在5年了没复发,想买重疾和医疗,有什么产品?""甲状腺癌全切术后5年未复发能否投保重疾险?""甲状腺癌全切术后5年未复发能否投保医疗险?"
"乳腺癌二期化疗结束两年了,想知道能买什么保险,保费大概多少?""乳腺癌二期化疗结束2年能否投保?""乳腺癌二期化疗结束2年可投保产品的保费是多少?"

执行方式: 每个拆解出的核心问题,对每个需要查询的产品分别调用一次脚本。即:查询次数 = 核心问题数 × 产品数。

产品名称范围:众民保·复发险(慢性白血病计划)众民保·复发险(急性白血病计划)众民保·中高端医疗险2026众民保·重疾险(免健告)众民保·复发险众民保·百万医疗险2025

根据用户疾病类型选择相关产品查询,例如:白血病相关优先查复发险系列;一般疾病查中高端/百万医疗。无法判断时,逐一查所有产品。

⚠️ 查询充分性校验(每轮查询后必须执行):

每次调用脚本或 web_search 后,必须评估返回结果是否足以回答用户的核心问题:

  • 结果充分:包含明确的核保结论(可投/除外/拒保)或相关保障/价格信息 → 继续下一步
  • 结果不足:返回内容模糊、未覆盖用户疾病场景、或仅有泛泛介绍无具体核保规则 → 必须补充查询

补充查询策略:

  1. 换一个更精准的问题描述重新调用脚本(调整疾病描述的细节,如加入分期、指标、治疗状态等)
  2. 若脚本连续两次结果仍不足,改用 web_search 补充搜索该疾病在对应产品的核保规则
  3. 若所有途径均无明确结论,在输出中如实说明"暂无明确核保规则,建议联系核保人员确认",不得凭训练知识捏造结论

其他保司产品(至少 2-3 款)→ 参考 references/other-products.md,用 web_search 搜索各产品针对该疾病的核保规则:

web_search("<产品名称> <疾病名称> 核保规则 带病投保")

Step 3:输出类比

将众民保系列与其他保司产品并列展示,客观呈现各家情况,不做优劣评价,让用户自行判断

  • 按产品逐一列出:核保结论(可投/除外/拒保)、保障范围、主要限制条件
  • 不使用"更好"、"推荐"、"建议选"等倾向性表述
  • 投保注意事项 + 健康告知填写指导
  • 附加合规免责声明(见 references/compliance-rules.md
  • 用户有意向时:

Comments

Loading comments...