Skill flagged — suspicious patterns detected

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

Veterinary Clinic Bundle

v1.0.1

VetClaw - 52个宠物医院/兽医诊所AI自动化技能套装。覆盖排班管理、病历记录、客户回访、经营分析等场景。

0· 74·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Suspicious
medium confidence
Purpose & Capability
The skill name/description (vet clinic automation) aligns with the included code and handlers (appointment, intake, emergency, inventory, etc.). However, the registry metadata declares no required env vars/credentials while the configuration files and code clearly expect multiple external service credentials (DEEPSEEK_API_KEY, SMS_API_KEY/SMS_API_SECRET, WECHAT_APP_ID/WECHAT_APP_SECRET, DB_PASSWORD, etc.). That omission is inconsistent and reduces transparency.
Instruction Scope
SKILL.md and README provide sensible setup steps (edit config/vet-config.yaml, set LLM provider, run installer). They instruct storing data locally or on your cloud. But runtime instructions and code cause the agent/server to read configuration and environment variables, persist PII/medical records into a local SQLite DB, and — if configured — transmit user data to external services (DeepSeek LLM, Aliyun SMS, WeChat, payment APIs). The docs do not explicitly warn that medical/personal data may be sent to third‑party LLMs or SMS/WeChat providers when those integrations are enabled.
Install Mechanism
There is no external installer that downloads code from unknown URLs; the bundle is code-included and contains a Python install script and a FastAPI app. No remote archive/extract/install URLs are present in the manifest. Installing means running included Python scripts and pip installing listed Python dependencies (requirements.txt).
!
Credentials
Although the registry lists no required env vars, the code and config reference multiple secret-bearing environment variables (DEEPSEEK_API_KEY, SMS_API_KEY/SMS_API_SECRET, WECHAT_APP_ID/WECHAT_APP_SECRET, DB_PASSWORD). Those are legitimate for the described integrations, but the metadata omission is a transparency problem. Providing these keys will cause user text (potentially PII or sensitive medical symptom descriptions) to be transmitted to third‑party services; that risk should be explicit. The number of secret/credential types is moderate but must be disclosed.
Persistence & Privilege
The skill does not request forced 'always' installation and does not attempt to modify other skills. It creates a local SQLite DB (./data/vetclaw.db) and writes skill files via the included installer script — normal for a local web app. There is no code attempting to alter other agent configs or escalate privileges.
What to consider before installing
This bundle appears to implement the described vet-clinic features, but be aware of these issues before installing: - Missing declared credentials: The package metadata says no env vars required, but the code/config expect several secrets (DEEPSEEK_API_KEY, SMS_API_KEY/SMS_API_SECRET, WECHAT_APP_ID/WECHAT_APP_SECRET, optional DB_PASSWORD). Ask the publisher to update the registry to list required env vars before providing keys. - Third‑party data flow: If you set an LLM provider API key (DeepSeek/Qwen/OpenAI) or enable SMS/WeChat, user messages and medical symptom details will be sent to those external services. Only use API keys for providers you trust and ensure that sending clinical or personal data to them complies with your privacy/legal requirements. - Run in an isolated/test environment first: The bundle creates a local SQLite DB and starts a web service. Test on a non-production host and inspect logs/traffic to ensure no unexpected outbound endpoints. - Review code for integration endpoints: Confirm the external endpoints (e.g., https://api.deepseek.com) are legitimate and acceptable for your data policy. If you need to avoid external LLMs, run with no API key so the code falls back to local/template behavior. - Ask for provenance: Source/homepage is 'unknown' but README references a GitHub repo and support email; verify the project repository and author identity before trusting production data. What would raise confidence: explicit registry declaration of required env vars, a published source repository (with commit history and maintainer info), and a privacy statement clarifying what data is sent to which third parties. If you want, I can: (1) list every point in code where an outbound network call is made, (2) extract all environment variable names the code reads, or (3) suggest minimal config settings to run the app offline (no external APIs).

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

latestvk97dn20zy0f9qg0yr3z3enc1qs83m0rr

License

MIT-0
Free to use, modify, and redistribute. No attribution required.

Comments