Skillv1.1.0

ClawScan security

HappyHorse 视频创作助手 · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

SuspiciousApr 28, 2026, 8:44 AM

Verdict: suspicious
Confidence: medium
Model: gpt-5-mini
Summary: The skill's functionality (calling Alibaba DashScope HappyHorse to generate videos) matches its description, but the package embeds a hard-coded API key and uses example/test hosts which are not declared as required — this is disproportionate and risky.
Guidance: This skill appears to do what it says (call Alibaba DashScope to generate videos), but it includes a hard-coded API key in both SKILL.md and the Python module and uses an example test image server IP. Before installing or running: - Do not rely on the embedded API key. Replace it with your own DashScope/Alibaba API key or require the skill read the key from a user-provided environment variable or prompt. - Treat the embedded key as compromised: it may be revoked, belong to someone else, or incur charges if abused. Avoid using it for production or sensitive data. - Inspect and avoid the example/test host (43.167.197.36) — it may be an unmanaged server. Prefer trusted image URLs and host your own media or use verified CDNs. - Be aware the skill will make outbound HTTP(S) requests (create tasks, poll status, download media). That can leak prompts, media URLs, and generated content to the API provider. If that is a privacy concern, do not use the skill. - If you trust the author and want to proceed: edit the code to require an environment variable (e.g., DASHSCOPE_API_KEY) instead of using DEFAULT_API_KEY, rotate any real credentials after testing, and review network egress policies and quota/cost implications for the DashScope service. If you want higher assurance, ask the author to remove embedded credentials, document the source of the test hosts, and switch examples to clearly labelled placeholders rather than live tokens/hosts.

Review Dimensions

Purpose & Capability: noteThe skill implements video generation against the DashScope / HappyHorse API and only requires python3, which matches the stated purpose. However, the SKILL.md and code include a hard-coded API key and test IPs/URLs that are not justified by the declared requirements (no credentials required).
Instruction Scope: noteRuntime instructions show exactly the API calls the skill will make and the expected workflow (create task, poll status, download result). This stays within the stated purpose. But instructions and examples embed a Bearer token and a test image server (43.167.197.36) and state the API key is '默认已配置（测试可用）', which implicitly encourages using an embedded/unknown credential and contacting external test hosts.
Install Mechanism: okNo install spec is provided (instruction-only plus a single Python module). Only python3 is required — minimal disk/write footprint and low install risk.
Credentials: concernThe skill does not declare any required environment variables, yet both SKILL.md and the Python module include a DEFAULT_API_KEY hard-coded in the repository. Embedding an API key is disproportionate: it exposes a credential in source, may be a third-party/author key you should not use, and could incur costs or privacy issues. The examples and tests also reference a raw IP image host (43.167.197.36), which may be an unmanaged endpoint.
Persistence & Privilege: okThe skill does not request persistent/global privileges (always:false), does not modify system-wide settings, and writes outputs to a workspace under /root/.openclaw — consistent with a user-level skill.