Go Stargazing Trip

• Outcome: Review
• Latest audit: May 29, 2026 · 9:29 PM UTC
• Version: 0.1.0

Overview

The skill’s stargazing trip-planning purpose is coherent, but its live weather fetch silently disables TLS certificate verification on retries, which can compromise recommendation integrity.

Review before installing if you rely on live weather. The planner’s main purpose looks legitimate, but the weather fetch should be changed to always use normal HTTPS certificate verification and fail clearly if secure retrieval fails.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 80% confidence
• Finding: The skill metadata declares no permissions, yet the implementation reportedly has file read, network, and shell capabilities. That mismatch is dangerous because callers and reviewers cannot accurately assess what the skill may do at runtime, and shell/network access materially expands the attack surface for command execution, data exfiltration, and unreviewed external interactions.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 92% confidence
• Finding: The documented purpose says this skill only performs multi-night route continuity recommendations, but the code reportedly also does single-night recommendation, live weather fetching, astronomical calculations, calendar conversion, and testing tooling. This description-behavior gap undermines trust boundaries and can cause the agent to invoke a much more capable skill than intended, increasing risk from hidden network activity, broader data processing, and execution of non-user-facing support code.

Intent-Code Divergence

High

• Confidence: 99% confidence
• Finding: The retry path explicitly sets `insecure = attempt > 0`, and `_fetch_json_via_urllib` uses `ssl._create_unverified_context()` when `insecure=True`, disabling TLS certificate validation on retries. This enables man-in-the-middle interception or response tampering from any network position able to interfere with the connection, which is especially risky because the tampered weather data directly affects route recommendations and users may rely on it for travel decisions.

Description-Behavior Mismatch

High

• Confidence: 96% confidence
• Finding: This skill's purpose is itinerary planning, but it contains network retry behavior that silently downgrades transport security for weather retrieval. That broadens the attack surface beyond the skill's core planning logic and allows attackers to inject forged weather payloads that can manipulate output recommendations, reliability, and potentially user safety during remote nighttime travel.

Missing User Warnings

High

• Confidence: 99% confidence
• Finding: The code disables certificate verification on retry without any user-facing warning, audit signal, or consent, so consumers cannot distinguish trusted weather data from data fetched over an unverifiable TLS session. Silent security downgrades are dangerous because they hide integrity failures and make compromised recommendations appear legitimate.

VirusTotal

64/64 vendors flagged this skill as clean.

Static analysis

No suspicious patterns detected.