Back to skill

Security audit

Geoskill: 河流形态变化

Security checks across malware telemetry and agentic risk

Overview

This skill performs river morphology analysis and has some disclosed network download and local caching behavior, but I found no artifact-backed deception, exfiltration, destructive action, or hidden high-impact authority.

Install only in environments where outbound access to Microsoft Planetary Computer and local geospatial output files are acceptable. For controlled or reproducible deployments, pin the dependencies and set an explicit output and cache directory so downloaded imagery can be reviewed and cleaned up.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (12)

Lp3

Medium
Category
MCP Least Privilege
Confidence
84% confidence
Finding
The skill documentation indicates output file creation and local caching behavior, but no explicit permissions are declared for file writes. Undeclared write capability weakens the trust boundary for users and orchestrators because the skill can modify the filesystem beyond what its manifest transparently communicates.

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding
The manifest presents a bounded local-analysis skill for provided water masks, but the documentation adds remote imagery discovery/download and synthetic data generation. This hidden expansion of capability changes the security model by introducing network access, untrusted external inputs, and potentially misleading outputs that users may not expect from the declared purpose.

Intent-Code Divergence

Medium
Confidence
92% confidence
Finding
The documentation contradicts the earlier local-mask processing model by stating that the skill can automatically download imagery from Microsoft Planetary Computer. Contradictory input models increase the chance of unintended network activity and processing of data from sources the user did not intend to trust.

Description-Behavior Mismatch

Medium
Confidence
89% confidence
Finding
The manifest describes a constrained morphology-analysis skill, but the added external data acquisition capability is omitted from the declared description. This creates a disclosure gap that can mislead policy engines and users about what resources the skill may access.

Context-Inappropriate Capability

Medium
Confidence
87% confidence
Finding
Automatic downloading from an external service is broader than the stated purpose of analyzing provided water body masks. Even if the source is legitimate, unnecessary network capability expands attack surface and can expose users to unexpected data handling, dependency, and provenance risks.

Description-Behavior Mismatch

Medium
Confidence
93% confidence
Finding
The skill advertises analysis of supplied water body masks, but it also contains hidden network-capable behavior that can search and download Sentinel-2 imagery from Planetary Computer when bbox/date arguments are provided. This expands the trust and attack surface beyond the stated purpose, can trigger unexpected outbound data access, and may violate least-surprise or environment restrictions in agent deployments.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The markdown advertises automatic remote downloads and local caching without warning users about network activity, storage side effects, or where cached data is retained. Lack of disclosure can cause privacy, compliance, and operational issues in restricted environments.

Unpinned Dependencies

Low
Category
Supply Chain
Content
geoskill-data-fetcher>=0.2.0

# Third-party dependencies used by this skill
numpy
rasterio
scipy
shapely
Confidence
91% confidence
Finding
numpy

Unpinned Dependencies

Low
Category
Supply Chain
Content
# Third-party dependencies used by this skill
numpy
rasterio
scipy
shapely
Confidence
91% confidence
Finding
rasterio

Unpinned Dependencies

Low
Category
Supply Chain
Content
# Third-party dependencies used by this skill
numpy
rasterio
scipy
shapely
Confidence
91% confidence
Finding
scipy

Unpinned Dependencies

Low
Category
Supply Chain
Content
numpy
rasterio
scipy
shapely
Confidence
90% confidence
Finding
shapely

Unpinned Dependencies

Low
Category
Supply Chain
Content
# Requirements for river-morphology-change
# Shared data-download library (Planetary Computer / NASA POWER / OSM)
geoskill-data-fetcher>=0.2.0

# Third-party dependencies used by this skill
numpy
Confidence
93% confidence
Finding
geoskill-data-fetcher>=0.2.0

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.