Security audit

hectorlee-momentum-position-advisor

Security checks across malware telemetry and agentic risk

Overview

This is a disclosed A-share momentum analysis skill that reads user-directed stock/portfolio inputs and outputs trading signals without executing trades or hiding data access.

Install only if you want a technical A-share momentum advisor. Review the broad triggers, the dependency on the local volume-price-screener skill, and the sector-resonance downgrade logic before relying on outputs; treat HOLD/REDUCE/SELL as informational signals that still require your own review.

SkillSpector

By NVIDIA

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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 (6)

Tp4

High

Category: MCP Tool Poisoning
Confidence: 76% confidence
Finding: The skill is presented as a momentum position advisor, but the documented behavior extends into market-wide scanning, local portfolio ingestion, stock search, cross-skill data fusion, and sector-level downgrade logic. This mismatch can cause users or orchestrators to grant trust, data access, or execution scope under narrower assumptions than the skill actually requires, increasing the chance of unintended data exposure or over-broad actions.

Description-Behavior Mismatch

Medium

Confidence: 93% confidence
Finding: The skill is described as a position-holding advisor, but the CLI also exposes stock search and full-market scanning. This expands the operational scope from portfolio advice into idea generation and broad market screening, which can violate least-functionality expectations and cause the agent to perform actions the user did not intend.

Description-Behavior Mismatch

Low

Confidence: 84% confidence
Finding: The code dynamically extends sys.path to import modules from another local skill directory under ~/.workbuddy/skills/volume-price-screener/scripts. This creates hidden cross-skill coupling and trust expansion: behavior now depends on external code outside this skill's declared boundary, increasing the chance of unexpected capabilities or execution of tampered local modules.

Vague Triggers

Medium

Confidence: 88% confidence
Finding: The trigger phrase set includes very generic investing intents such as '减仓' and '趋势跟踪', which can cause the platform to invoke this skill for broad financial-advice queries outside its narrowly described momentum/position-management scope. In a finance context, overbroad routing is risky because users may receive deterministic trading guidance from a specialized technical-analysis skill when their question actually requires broader suitability, risk, or portfolio context.

Vague Triggers

Medium

Confidence: 92% confidence
Finding: Multiple manifest triggers are ambiguous and lack boundaries around asset class, market, and task type, increasing the chance that unrelated or higher-stakes investment questions are matched to this skill. Because the skill is framed as a position advisor, accidental invocation can lead to overconfident buy/sell/hold outputs in situations where the user did not intend to use a pure momentum strategy.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: This code automatically applies sector-wide score downgrades and can escalate a position from hold/reduce to sell based solely on peer signals in the same inferred sector, with no built-in user acknowledgment, confirmation, or explicit opt-in at the decision point. In a trading advisory skill, that is dangerous because sector classification is heuristic and the logic can force materially different actions across holdings, increasing the risk of unintended liquidation or systematic overreaction during volatility.

VirusTotal

59/59 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.