Security audit

Quantitative Research

Security checks for vulnerabilities and agentic risk

Overview

This skill is a finance research guidance package with broad routing terms, but it does not request credentials, persistence, trade execution, or hidden system access.

Install only if you want quantitative finance research assistance. Treat its strategy code and thresholds as educational examples, verify any market data and assumptions independently, and do not rely on it as financial advice or as authority to trade without explicit user-controlled review.

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

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

Findings (5)

Vague Triggers

Medium

Confidence: 94% confidence
Finding: The trigger description is overly broad and ambiguous, causing the skill to activate for a wide range of finance-related prompts without clear boundaries. In an agent system, this can misroute user requests to a high-authority domain skill, increasing the chance of inappropriate guidance, overreach beyond intended scope, or unintended instruction precedence over safer/more relevant skills.

Vague Triggers

Medium

Confidence: 97% confidence
Finding: The detection rule `alpha|signal|predict|edge` is overly broad for an auto-invocation mechanism because terms like 'signal', 'predict', and especially 'edge' appear in many unrelated technical and general contexts. This can cause the skill to trigger unexpectedly, routing user requests into a powerful quant-research workflow when the user did not intend it.

Vague Triggers

Medium

Confidence: 95% confidence
Finding: The detection pattern `factor|exposure|beta|Fama.*French|Barra` contains generic terms like 'factor', 'exposure', and 'beta' that are common across software, statistics, science, and product discussions. In a skill router, this increases the chance of accidental invocation and misclassification of unrelated requests.

Vague Triggers

Medium

Confidence: 96% confidence
Finding: The detection rule `regime|market.*state|trend.*ranging|volatility.*cluster` includes broad terms like 'regime' and potentially generic phrasing around trends and volatility. These words are used in many non-financial domains, so the skill may activate on unrelated discussions about systems, policies, ML, or weather-like volatility concepts.

Vague Triggers

Medium

Confidence: 97% confidence
Finding: The anti-pattern triggers include very generic words like `download`, `historical`, `future`, `position`, `trade`, `execute`, `recent`, and similar terms that commonly appear in ordinary coding and research conversations. If these are used for skill behavior, warning generation, or routing, they can produce noisy matches and unintended activations that degrade trust and can redirect workflows incorrectly.

Static analysis

No suspicious patterns detected.