Geoskill: 可再生能源场址选择

• Outcome: Review
• Latest audit: Jul 30, 2026 · 3:49 PM UTC
• Version: 1.0.0

Overview

The skill performs a coherent renewable-energy GIS workflow, but its automatic third-party data downloads and persistent caching are under-disclosed enough that users should review it before installing.

Install only if you are comfortable with bbox/AOI and date-range queries being sent to NASA POWER and the Planetary Computer, and with downloaded data being stored in the output directory or cache. Prefer local raster inputs for sensitive projects, set an explicit cache/output directory, and consider pinning dependencies before use in regulated or reproducible environments.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection

Findings (11)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 83% confidence
• Finding: The skill documents writing multiple output files but does not declare any corresponding permissions, which creates a transparency and governance gap. Undeclared file-write behavior can lead to unexpected local modifications, especially in agent environments where permission declarations are used to bound side effects.

Tp4

High

• Category: MCP Tool Poisoning
• Confidence: 95% confidence
• Finding: The declared purpose presents the skill as a local GIS suitability tool, but the documentation indicates additional behaviors including remote data retrieval, DEM search/download, synthetic data generation, and provenance export. This mismatch is dangerous because users and orchestrators may approve or invoke the skill under incorrect assumptions about network access, data origin, and side effects.

Description-Behavior Mismatch

Medium

• Confidence: 90% confidence
• Finding: The main documentation frames the skill as operating on local rasters, but a later section adds automatic remote download behavior from NASA POWER. Hidden expansion from local-only processing to network-enabled operation increases the risk of unanticipated data transfer, policy violations, and incorrect trust assumptions in restricted environments.

Description-Behavior Mismatch

Medium

• Confidence: 84% confidence
• Finding: The skill silently expands from local raster analysis into network-enabled data acquisition, downloading from NASA POWER and the Planetary Computer and writing results to disk. In an agent environment, undeclared outbound network access and side-effecting downloads can violate trust boundaries, leak user AOI/date metadata to third parties, and introduce supply-chain and reproducibility risks.

Context-Inappropriate Capability

Medium

• Confidence: 86% confidence
• Finding: The external data-harvesting capability is broader than expected for a site-selection analyzer and causes the tool to contact third-party services based on user-provided AOI and dates. That can expose sensitive project locations, create compliance concerns in restricted environments, and pull unreviewed external content into the workflow.

Missing User Warnings

Medium

• Confidence: 91% confidence
• Finding: The documentation describes automatic network downloads and local caching without an explicit warning that invoking the skill may contact external services and transmit user-supplied AOI or bounding-box information. In regulated or sandboxed environments, undisclosed network access is a meaningful security and privacy risk.

Missing User Warnings

Medium

• Confidence: 90% confidence
• Finding: The code performs remote downloads and writes the resulting CSV and possibly DEM assets to disk without a strong, explicit user warning or consent gate. In agent-driven execution, hidden network and file-write side effects are dangerous because they can surprise operators, disclose sensitive query parameters, and create unwanted artifacts on the filesystem.

Unpinned Dependencies

Low

• Category: Supply Chain

Content

geoskill-data-fetcher>=0.2.0

# Third-party dependencies used by this skill
numpy
rasterio
shapely

• Confidence: 97% confidence
• Finding: numpy

Unpinned Dependencies

Low

• Category: Supply Chain

Content

# Third-party dependencies used by this skill
numpy
rasterio
shapely

• Confidence: 97% confidence
• Finding: rasterio

Unpinned Dependencies

Low

• Category: Supply Chain

Content

# Third-party dependencies used by this skill
numpy
rasterio
shapely

• Confidence: 97% confidence
• Finding: shapely

Unpinned Dependencies

Low

• Category: Supply Chain

Content

# Requirements for renewable-energy-site-selection
# Shared data-download library (Planetary Computer / NASA POWER / OSM)
geoskill-data-fetcher>=0.2.0

# Third-party dependencies used by this skill
numpy

• Confidence: 99% confidence
• Finding: geoskill-data-fetcher>=0.2.0

VirusTotal

65/65 vendors flagged this skill as clean.

Static analysis

Detected: suspicious.dynamic_code_execution

Dynamic code execution detected.

Critical

• Code: suspicious.dynamic_code_execution
• Location: scripts/renewable_energy_site_selection.py:94