Security audit
Security checks for vulnerabilities and agentic risk
ClawScan marks this skill benign with medium confidence: the reported issues concern an unpinned install command, unclear tool scope, and possible reliability problems, while the scan was partial because some references were unresolved.
Consider the unpinned install command, unclear tool scope, and reported calculation and evidence-rating issues before installing. The verdict is benign, not risk-free, and the unresolved references mean the evidence is incomplete.
The installation command invokes an unpinned npm package through npx, allowing the resolved package version to change and execute code with the installing user's privileges. This is a supply-chain exposure rather than an MCP-server issue; it requires the user to run the documented installation command.
Suspicious Unicode normalization or mixed-script content
Without declared permissions the skill's intent is opaque and cannot be validated.
The delivery instructions explicitly allow paid custom reports to receive L2 status, contradicting the policy that reserves L2 for testing without commercial relationships and assigns paid custom reports to L3c. This ambiguity can launder commercially influenced evidence into a stronger trust tier, weakening recommendation safeguards in a workflow that includes safety-critical products, although conflict labeling and cross-check requirements reduce the exposure.
Suspicious Unicode normalization or mixed-script content
Suspicious Unicode normalization or mixed-script content
Suspicious Unicode normalization or mixed-script content
subprocess module calls execute external commands. Without careful input validation, this enables command injection.
def fetch(url: str) -> str:
"""用 curl 取页面，避开 python urllib 的 SSL 握手问题。"""
proc = subprocess.run(
["curl", "-s", "--compressed", "-A", UA,
"-H", "Accept-Language: zh-CN,zh;q=0.9",
"-m", "30", "-L", url],
Suspicious Unicode normalization or mixed-script content
The exposure calculation places the uninsured duration at the beginning of ownership rather than after the warranty expires. With a positive discount rate, this overstates repair present value and can distort EAC, comparative rankings, and risk thresholds; zero-discount cases mask the defect. This matters because the skill presents these calculations as quantitative support for expensive or safety-relevant purchases.
These terms can occur in unrelated requests, such as document comparisons, games, or financial analysis. Although the description presents a product-selection skill, it does not explicitly constrain these trigger matches to shopping requests or provide exclusions, risking unintended activation of the research workflow.