Security audit

yanmu-comps-model

Security checks across malware telemetry and agentic risk

Overview

This skill is a stock comparable-valuation helper that uses a bundled dataset and disclosed live price lookups, with no evidence of hidden persistence, credential access, or destructive behavior.

Install only if you are comfortable with the skill making outbound Sina Finance quote requests for tickers you analyze and writing chart images to the chosen output directory. Treat the generated financial analysis as informational, not investment advice.

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: 90% confidence
Finding: The skill documentation indicates automatic network access to Sina Finance for live prices, but no permissions are declared. Undeclared network capability weakens user consent and platform control, and could allow unexpected outbound requests when the skill is invoked.

Tp4

High

Category: MCP Tool Poisoning
Confidence: 93% confidence
Finding: The skill presents itself primarily as a comparable-company valuation tool based on an internal database, but also performs live external data retrieval and recalculates valuation metrics from remote content. This behavior mismatch is security-relevant because users and policy systems may authorize the skill under incomplete assumptions about data flows and runtime capabilities.

Description-Behavior Mismatch

Medium

Confidence: 92% confidence
Finding: The script performs undisclosed outbound network access to Sina Finance to fetch live market data, which expands the skill's behavior beyond local comps analysis and creates data-flow, privacy, and reliability risks. In an agent environment, hidden external requests can leak user interest in specific tickers, introduce untrusted third-party content into outputs, and bypass expectations that the skill operates only on bundled data.

Missing User Warnings

Low

Confidence: 82% confidence
Finding: The markdown states that live prices are fetched automatically from Sina Finance but does not warn about network use, third-party data transfer, or trust in remote responses. Even if the transferred data is limited, silent outbound access reduces transparency and can surprise users in restricted or sensitive environments.

VirusTotal

53/53 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.