Security audit

盘前简报Pro

Security checks across malware telemetry and agentic risk

Overview

This is an offline educational stock-analysis demo that reads a local config file and writes local reports, with a privacy caveat around portfolio-style sample fields.

Install only if you want a local offline teaching demo. Keep the sample holdings or protect/delete config.json and logs if you enter real portfolio information, and run it as a normal user rather than using the /root cron path shown in the example.

SkillSpector

By NVIDIA

Vulnerability Patterns

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
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (2)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 81% confidence
Finding: The skill documentation instructs users to copy files into the OpenClaw workspace and run a Python script, and it also describes local log storage and configuration files. That means the skill clearly relies on file read/write behavior, but no permissions are declared, creating a transparency and policy gap that can hide filesystem access from users and reviewers.

Intent-Code Divergence

Medium

Confidence: 74% confidence
Finding: The document claims the system is purely simulated and does not read real data, yet the example configuration requests real holdings, share counts, and cost prices. This contradiction can mislead users into supplying sensitive financial information under the belief that the skill is harmless and purely educational, increasing privacy and trust risk even in an allegedly offline context.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.