Back to skill
Skillv0.1.0
ClawScan security
科普智创审校助手 · ClawHub's context-aware review of the artifact, metadata, and declared behavior.
Scanner verdict
BenignApr 5, 2026, 5:30 AM
- Verdict
- benign
- Confidence
- high
- Model
- gpt-5-mini
- Summary
- The skill's instructions, required resources, and scope are coherent with its stated purpose (collecting topics from kepuchina.cn, drafting and fact-checking popular-science articles); it is instruction-only, requests no credentials, and has no install steps.
- Guidance
- This skill appears internally consistent, but before installing verify: (1) your agent environment actually provides the referenced tools (tavily_search/tavily_extract) or you understand how those are implemented; (2) you are comfortable allowing the agent to fetch and parse external pages (network access) and to store or display any scraped content; (3) scraping kepuchina.cn complies with that site's terms and copyright; and (4) confirm the agent will not forward scraped content to third-party endpoints you don't control. If you need stronger guarantees, request the skill author to document the 'tavily' tool, its data handling, and any caching/retention behavior.
Review Dimensions
- Purpose & Capability
- noteThe name/description (hotspot collection, article drafting, and review) matches the steps in SKILL.md. The tools referenced (curl, tavily_search/tavily_extract) are appropriate for web collection and verification. Minor mismatch: identifying the 'top 10 by reading/clicks' may require analytics not publicly available on the site, so the skill assumes popularity metrics are inferable from public pages (or from Tavily).
- Instruction Scope
- noteInstructions direct the agent to fetch and parse kepuchina.cn pages and to run searches (tavily_search/extract) to verify claims and DOIs—these are in-scope for the stated monitoring and fact-checking purpose. The SKILL.md does assume existence of the 'tavily' tool/API without declaring where it comes from; it also gives a raw curl example which will perform network fetches. There is no instruction to read unrelated local files, request unrelated credentials, or send data to unknown external endpoints.
- Install Mechanism
- okInstruction-only skill with no install spec and no code files — lowest-risk from installation standpoint.
- Credentials
- okNo environment variables, credentials, or config paths are requested. All required actions (web fetch and search) can be performed without additional secrets, which is proportionate to the described functionality.
- Persistence & Privilege
- okalways:false and no install hooks or requests to modify agent/system configuration. The skill does not request permanent presence or elevated privileges.
