Security audit

dca-decision-calculator

Security checks across malware telemetry and agentic risk

Overview

This finance skill is mostly transparent, but its very broad triggers can activate investment and currency guidance outside a clearly requested calculation workflow.

Review before installing if you discuss finances broadly in the same agent environment. The skill does not appear to run code or access accounts, but it may activate on ordinary finance wording and produce advice-like DCA, exchange, holding, or reduction suggestions; use it only when you explicitly want that workflow and verify calculations independently.

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 (21)

Intent-Code Divergence

Medium

Confidence: 91% confidence
Finding: The skill states an 'iron law' that no operational advice should be given before required data is collected, but later sections allow entry into diagnosis/strategy modes with looser conditions. This inconsistency can cause the agent to provide financial guidance from incomplete inputs, increasing the risk of misleading or unsuitable recommendations.

Description-Behavior Mismatch

Medium

Confidence: 89% confidence
Finding: The manifest presents the skill as a narrow calculator, but the body expands into valuation analysis, portfolio health assessment, and buy/hold/reduce guidance. This scope drift is dangerous because a skill invoked as a simple calculator may instead deliver broader financial-advice behavior not clearly disclosed or constrained.

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The trigger list includes very common finance terms such as '投资', '成本', '指数', and '汇率', which can cause the skill to activate during ordinary financial discussions unrelated to this specific calculator. In an agent environment, unintended invocation can expose user financial context to the skill and lead to unsolicited investment guidance, making misrouting and over-collection of sensitive context a real security and safety concern.

Vague Triggers

High

Confidence: 95% confidence
Finding: The trigger list includes many common finance terms that are likely to appear in ordinary conversation, making accidental invocation probable. In a finance context, unintended activation is more dangerous because the skill may steer users into investment or currency decisions when they were not explicitly requesting this workflow.

Vague Triggers

High

Confidence: 97% confidence
Finding: The manifest enumerates ambiguous one- or two-word triggers like '基金', '汇率', and '买入' without contextual constraints. Such broad matching can hijack unrelated financial chats and route them into a calculation/advice flow, increasing the chance of inappropriate or unsolicited financial guidance.

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The mode-entry table does not clearly define precedence, minimum data requirements, or when the skill should remain inactive. Ambiguous routing can lead the model to choose a more advisory mode than intended and produce outputs beyond the user's explicit request or available data.

Overly Broad Trigger

Low

Category: Trigger Abuse
Confidence: 83% confidence
Finding: The trigger '定投' is very short and broadly used in investment discussions, so it can activate the skill even when the user is only mentioning the concept generally. In this finance skill, accidental invocation can still lead to unsolicited calculation or strategy guidance.

Overly Broad Trigger

Low

Category: Trigger Abuse
Confidence: 82% confidence
Finding: The trigger '基金' is too generic and appears in many unrelated financial conversations. This broadness raises the chance that the skill is invoked when the user did not request a DCA decision workflow.

Overly Broad Trigger

Low

Category: Trigger Abuse
Confidence: 82% confidence
Finding: The trigger '指数' is a common financial term with many non-DCA uses. Its use as a standalone activator can cause over-triggering and redirect ordinary market discussion into this skill.

Overly Broad Trigger

Low

Category: Trigger Abuse
Confidence: 84% confidence
Finding: The trigger '加仓' is short and may appear in broad investment chats without implying the user wants this specific left-side DCA strategy. Inadvertent activation can produce recommendation-like outputs that exceed user intent.

Overly Broad Trigger

Low

Category: Trigger Abuse
Confidence: 84% confidence
Finding: The trigger '换汇' is generic and could appear in routine currency conversations unrelated to this skill's strategy. Because the skill can output timing-style suggestions, accidental activation carries practical user-harm risk.

Overly Broad Trigger

Low

Category: Trigger Abuse
Confidence: 82% confidence
Finding: The trigger '汇率' is extremely common in benign discussion and does not uniquely identify this skill's purpose. This can lead to unsolicited exchange-cost suggestions or strategy framing.

Overly Broad Trigger

Low

Category: Trigger Abuse
Confidence: 81% confidence
Finding: The trigger '投资' is overly broad and may match a wide range of general conversations. In a financial skill that can suggest actions, this increases the chance of unrequested guidance.

Overly Broad Trigger

Low

Category: Trigger Abuse
Confidence: 83% confidence
Finding: The trigger '买入' is a common action word that appears in many contexts, not just this DCA calculator workflow. As a standalone trigger, it can cause the skill to activate on casual discussion and emit recommendation-like outputs.

Overly Broad Trigger

Low

Category: Trigger Abuse
Confidence: 83% confidence
Finding: The trigger '持仓' is frequently used in portfolio discussions and does not uniquely map to the skill's intended calculation. This broadness can pull in general portfolio conversations and generate unintended diagnostic advice.

Overly Broad Trigger

Low

Category: Trigger Abuse
Confidence: 82% confidence
Finding: The trigger '成本' is too generic and can appear in many unrelated contexts, including non-financial ones. Using it as a standalone activator creates unnecessary risk of spurious skill selection.

Overly Broad Trigger

Low

Category: Trigger Abuse
Confidence: 80% confidence
Finding: The trigger '跌了' is casual, ambiguous language common in market chat. This can over-activate the skill on conversational remarks rather than explicit requests for DCA analysis.

Overly Broad Trigger

Low

Category: Trigger Abuse
Confidence: 80% confidence
Finding: The trigger '涨了' is similarly ambiguous and common in ordinary discussion. Its use can cause the skill to activate even when the user is not seeking strategy outputs.

Overly Broad Trigger

Low

Category: Trigger Abuse
Confidence: 83% confidence
Finding: The trigger '跌幅' is short and broadly used in market commentary. As a standalone trigger, it is likely to match users who want information, not an investment-action calculator.

Overly Broad Trigger

Low

Category: Trigger Abuse
Confidence: 78% confidence
Finding: The trigger '缩水' is colloquial and can be used in many contexts, including general complaints about losses. While lower risk than some generic finance terms, it is still too ambiguous as a sole activator for a financial decision skill.

Overly Broad Trigger

Low

Category: Trigger Abuse
Confidence: 82% confidence
Finding: The trigger '估值' is a common analysis term and not specific to this skill's left-side DCA calculations. It can attract general valuation discussions into a workflow that produces broader advisory content.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.