ClawHub
Back to skill

Security audit

Campus Morning Brief / 校园晨报

Security checks across malware telemetry and agentic risk

Overview

The skill is a disclosed student morning-brief workflow with expected profile use, current-information research, and local rendering outputs, with no evidence of hidden access, exfiltration, destructive behavior, or automatic delivery.

Before installing, understand that this skill may prompt your agent to use a student profile and browse for current weather and news. Use only the minimum student information needed, avoid sensitive reminders, review sources and generated files, and require explicit approval before anything is sent, posted, or printed.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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 (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
87% confidence
Finding
The skill explicitly instructs reading external profile/config files and writing generated outputs such as brief.json, morning-brief.md, morning-brief.html, and metadata.json, but no declared permissions accompany those capabilities. Undeclared file access undermines least-privilege review and can enable unintended access to local data or writes to unsafe locations, especially because the workflow allows an externally supplied output directory.

Tp4

High
Category
MCP Tool Poisoning
Confidence
81% confidence
Finding
The skill description promises a personalized, source-backed current-events brief, while the documented behavior primarily formats input, reads/writes files, and invokes a renderer. This mismatch is dangerous because users or automation may trust the skill to fetch and verify current weather/news when it may instead emit stale, unverified, or user-supplied content and write artifacts to disk without that behavior being clearly disclosed.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The promotion copy explicitly markets personalized briefs built from a student's city, age, interests, and reminders, which are all personal data elements and may include data about minors. Because this is student-focused and framed for educators and parents, omitting any privacy, consent, minimization, or retention guidance increases the risk that downstream users will collect or process children's data unsafely or in ways that violate policy or regulation.

Natural-Language Policy Violations

Medium
Confidence
93% confidence
Finding
The guideline requires output in the configured profile language, which can override an explicit or situational user preference if no opt-in or override path exists. This creates a privacy and autonomy issue because it relies on persistent profile data and may disclose or enforce inferred user attributes without current-session consent.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

Detected: suspicious.dynamic_code_execution

Dynamic code execution detected.

Critical
Code
suspicious.dynamic_code_execution
Location
tests/test_render_brief.py:14