Back to skill

Security audit

China Options Pricing

Security checks for vulnerabilities and agentic risk

Overview

The skill does not show malicious behavior, but it should be reviewed because its financial calculations and claimed strategy features are materially unreliable.

Review this skill before installing for real use. It has no evident credential, persistence, or exfiltration behavior, but its pricing math should not be relied on for trading or risk decisions until the normal CDF is fixed, examples are recalculated, and incomplete strategy-analysis claims are removed or implemented.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (1)

T09 · Insecure Skill Coding Practices

Error
Location
SKILL.md:64
Finding

Incorrect Normal CDF Implementation Corrupts Financial Pricing and Risk Calculations

Content
View full analysis
6: return 1.0 a1, a2, a3, a4, a5 = 0.254829592, -0.284496736, 1.421413741, -1.453152027, 1.061405429 p = 0.3275911 sign = 1.0 if x >= 0 else -1.0 x_abs = abs(x) t = 1.0 / (1.0 + p * x_abs) y = 1.0 - (((((a5 * t + a4) * t) + a3) * t + a2) * t + a1) * t * math.exp(-x_abs * x_abs / 2) return y if sign > 0 else 1.0 - y ``` ### Technical Analysis The coefficients in this implementation are intended for an approximation of the error function, `erf(x)`, rather than direct evaluation of the standard normal cumulative distribution function. The standard relationship is: ```text N(x) = 0.5 × (1 + erf(x / sqrt(2))) ``` The implementation does not apply the required `x / sqrt(2)` transformation or the final `0.5 × (1 + ...)` conversion. Consequently, it produces fundamentally incorrect results. For example, the standard normal CDF must satisfy `N(0) = 0.5`, whereas this implementation returns a value approximately equal to zero. This function is subsequently used by the documented Black–Scholes pricing, implied-volatility, and Greek calculations. The defect therefore propagates into option prices, Delta, Theta, Rho, and iterative implied-volatility results. Because the Skill presents those outputs as actionable financial analysis, this is a material integrity flaw. ### Attack Path 1. A user asks the agent to price an option, calculate implied volatility, or calculate Greeks. 2. The agent uses or reproduces the documented `norm_cdf` implementation. 3. The malformed CDF returns incorrect probabilities. 4. Those values are consumed by the Black–Scholes and risk calculations. 5. ...[truncated 1014 chars]
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
85% confidence
Finding

The title and introductory description are presented as Chinese-focused output behavior, and the quick-start section explicitly frames usage in Chinese without offering any language choice. Under the policy, forcing a specific language or locale without opt-in is a natural-language policy violation unless the constraint is clearly documented and justified.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill markets coverage of Chinese commodity options, many of which are explicitly noted later as American-style, but the implemented pricing examples only use European Black-Scholes formulas. This can mislead downstream users or agents into treating materially incorrect prices and Greeks as authoritative, creating financial decision risk in a trading context.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill claims multi-leg strategy P&L analysis, but the documented implementation leaves the main analysis function as a stub. In an agent setting, this can cause users or orchestrators to rely on nonexistent or partial functionality, leading to silent misanalysis of risk, payoff ranges, or break-even points.

Content

No source excerpt is available for this finding.