健康管理
ReviewAudited by ClawScan on May 10, 2026.
Overview
This health skill has a coherent purpose, but it handles very sensitive medical data, uses opaque API credentials, and has unsafe HTML report templating that should be reviewed before use.
Before installing, confirm you trust the publisher and external domains, because the skill asks for personal medical details and sends them to a backend model. Do not treat the AI report as a diagnosis. The publisher should remove embedded API keys, clearly document credential handling, and sanitize generated HTML reports.
Findings (4)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Your name, demographics, medical history, medications, allergies, lab reports, and conversation context may be sent to the external health-model provider.
The skill explicitly sends user-provided medical information and conversation context to an external backend model. This matches the health-assessment purpose, but the data is highly sensitive.
将这些信息携带上下文，通过指定的 API 接口发送给后端的医疗大模型（cyzh-cfc）
Use only if you trust the provider and are comfortable sharing this medical information; avoid submitting unnecessary identifiers or unrelated private details.
API requests may be made under a shared or embedded provider credential rather than a clearly scoped user-controlled credential.
A real-looking API key is distributed in the reference documentation, while registry metadata declares no credentials and the main instructions say the key is dynamically fetched. This makes the credential boundary and provenance unclear.
| API Key | `fastgpt-aIOCbwavMirdBk08oHQyDxudE56DBnksVXUL4m8c50CuPnTNGqK5yX7Ykm` |
The publisher should remove hardcoded keys, declare required credentials or token sources, and document which account or service identity is used.
A malicious or malformed health-report field could become executable content in the generated local HTML report.
Template values are inserted into the HTML report by raw string replacement without HTML escaping. If user input or API output contains HTML/JavaScript and the report is opened in a browser, injected script could run.
for key, value in data.items():
placeholder = f"{{{{{key}}}}}"
result = result.replace(placeholder, str(value))Escape all user- and API-supplied values before placing them into HTML, and avoid automatically opening generated reports unless the user explicitly requests it.
The skill may fail in environments without curl, and its runtime authentication depends on an external token endpoint.
The code requires curl at runtime to retrieve a token, but the registry says there are no required binaries. The behavior is disclosed in SKILL.md and is related to the API call, but the dependency and token source are under-declared.
["curl", "-s", "--max-time", "10", "https://jiyinjia.jinbaisen.com/!token?key=skill_jk"]
Declare the curl/runtime dependency or use the existing HTTP client library, and document the token endpoint and trust boundary clearly.