ClawHub
Back to skill

Security audit

Stanley Druckenmiller Workflow

Security checks across malware telemetry and agentic risk

Overview

This is a coherent public-market analysis skill that fetches financial data and caches snapshots locally, with the sensitive behavior disclosed and aligned to its purpose.

Install only if you are comfortable with a finance skill fetching public market data from third-party sources, optionally reading a FRED-specific API key, and storing local market-data cache files. Treat its outputs as research context, not investment advice or automated trading instructions.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
89% confidence
Finding
The skill declares no permissions, yet the analyzed behavior indicates access to environment variables, filesystem read/write, and network resources. That mismatch is dangerous because reviewers and users may trust the manifest boundary while the skill can actually fetch external data, persist local state, and potentially access sensitive runtime context without explicit disclosure.

Tp4

High
Category
MCP Tool Poisoning
Confidence
94% confidence
Finding
The stated purpose is a natural-language market workflow, but the observed behavior includes external data collection, cache persistence, raw JSON/panel generation, and CLI-style file output. This creates a trust-boundary problem: operators may approve a low-risk presentation skill while it actually behaves like a data-ingestion and local-state-processing tool, increasing the chance of unreviewed data exfiltration, unintended persistence, and misuse of host resources.

Vague Triggers

Medium
Confidence
88% confidence
Finding
The trigger list includes broad natural-language phrases such as 'macro update', 'how should I view today', and 'should I buy/sell' that can plausibly appear in ordinary conversation. In an agentic environment, this increases the chance of unintended skill activation, causing the skill to run in the wrong context and produce financial-analysis output when the user did not explicitly request this workflow.

Vague Triggers

Medium
Confidence
83% confidence
Finding
The invocation cues are broad enough to trigger on generic portfolio guidance, market-priority, and translation-style requests, which can cause the skill to activate outside narrow intended contexts. In a finance-oriented skill, overly broad routing increases the risk of unsolicited investment-style guidance, accidental overreach, and unexpected execution of the skill's wider data-access behavior.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.