Security audit

Portfolio Risk Analyzer

Security checks for vulnerabilities and agentic risk

Overview

This is a plain markdown skill for portfolio risk analysis; it handles sensitive financial information and uses web searches, but those behaviors fit its stated purpose and it does not install code, persist itself, or access accounts.

Install only if you are comfortable sharing portfolio details with your agent. Avoid providing account numbers, broker logins, personal identifiers, or unnecessary exact values, and prefer ticker-only external lookups. Treat rebalancing, tax-loss harvesting, and risk outputs as informational and verify them with trusted market data or a qualified financial professional before acting.

A.I.G

Vulnerability Patterns

Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (5)

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The skill states that the agent will look up current prices using web search, but it does not warn users that portfolio holdings and related financial context may be transmitted to external services. Because portfolio composition is sensitive financial data, silent transmission or indirect disclosure during lookups creates a meaningful privacy and confidentiality risk.

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The README presents several broad invocation examples like "What's my portfolio risk?" and "How should I rebalance?" without any trigger-scope constraints. In an agent ecosystem, these ambiguous phrases can cause the skill to activate during normal conversation, increasing the chance of unintended financial analysis, external data fetching, or disclosure of portfolio information.

Vague Triggers

Medium

Confidence: 89% confidence
Finding: The example command "What's my Value at Risk?" is generic enough to overlap with ordinary financial conversation, making accidental or opportunistic invocation more likely in systems that route on natural-language similarity. In this skill’s context, unintended activation could cause the agent to process sensitive portfolio details and perform web lookups without the user clearly intending to invoke this specific skill.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The skill solicits and processes detailed portfolio holdings, position sizes, cost basis, and cash balances, which are highly sensitive financial data. Without an explicit warning about sensitivity, data minimization guidance, or handling restrictions, users may disclose more information than intended and the system may process regulated or privacy-sensitive information without appropriate consent framing.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The skill explicitly directs the agent to perform web searches for current prices, volatility, beta, dividend yield, and rates without warning the user that portfolio symbols and analysis context may be sent to external providers. This can leak sensitive investment interests or holdings metadata to third parties and may cause users to rely on unverified external market data without understanding provenance or privacy implications.

Static analysis

No suspicious patterns detected.