Security audit

天气查询免费版

Security checks across malware telemetry and agentic risk

Overview

The skill appears to be a weather helper, but its instructions are broader than that purpose and request command/write authority, so it should be reviewed before installation.

Install only if you are comfortable with an agent using network weather APIs for your queried locations. Review or tighten the skill so it activates only for weather, forecast, air-quality, alert, and travel-weather requests, and avoid granting write access unless you explicitly want local cache/config files created.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Description-Behavior Mismatch

Medium

Confidence: 90% confidence
Finding: The skill is presented as a simple free weather-query tool, but the documentation adds generic create/query/export, modify/reset/import, and save/convert operations that expand the apparent authority of the skill well beyond its stated purpose. In an agent setting with read/exec/write tools enabled, this ambiguity can cause the agent to perform file writes, imports, resets, or exports unrelated to weather retrieval, increasing the risk of unintended state changes or data exfiltration.

Intent-Code Divergence

Medium

Confidence: 87% confidence
Finding: The trigger conditions say to use the skill for generic data analysis, report generation, statistics, and visualization, which is materially broader than a weather-query assistant. Overbroad routing criteria can cause an agent to invoke this skill for unrelated tasks, increasing the chance it uses exec/write capabilities or external API calls in contexts where the user did not intend weather-specific processing.

Vague Triggers

High

Confidence: 91% confidence
Finding: These trigger conditions are broad enough to match many unrelated requests, creating a skill-selection vulnerability. Because the skill advertises exec/write capabilities and external-network use, incorrect activation can lead to unintended command execution, file operations, or external data disclosure when a user only asked for general analysis or reporting.

Vague Triggers

Medium

Confidence: 80% confidence
Finding: The keyword coverage section enumerates a very broad set of loosely bounded phrases and asserts all corresponding scenarios are supported, without defining clear activation limits. In agent ecosystems, vague keyword-based coverage increases accidental invocation and can steer unrelated prompts into a skill that has command execution and write access available.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The skill does not clearly warn users that location queries and related weather requests are sent to third-party online APIs. Even though weather lookups seem low risk, location data can still be sensitive, and silent transmission to external services undermines informed consent and can create privacy/compliance issues.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.