Insurance Actuarial Python

• Outcome: Review
• Latest audit: May 28, 2026 · 1:57 AM UTC
• Version: 0.3.3

Overview

The skill is advertised as actuarial interest-rate modeling, but its own files steer the agent toward stock/crypto trading workflows and persistent setup, so it needs review before installation.

Install only if you intend to review and use it as a mixed financial/trading artifact, not just an actuarial modeling helper. Do not connect broker accounts, paid data-provider credentials, wallets, or live trading access without explicit per-action approval and a separate isolated Python environment.

SkillSpector

Vulnerability Patterns

• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration

Findings (14)

Description-Behavior Mismatch

High

• Confidence: 97% confidence
• Finding: The skill is presented as an actuarial interest-rate analysis tool, but its pipeline, prompts, and execution semantics pivot into equity/crypto data collection, backtesting, and live trading execution. This scope mismatch can mislead an orchestrator or user into invoking trading behavior under a benign analytics label, increasing the risk of unauthorized or unintended financial actions.

Intent-Code Divergence

Medium

• Confidence: 93% confidence
• Finding: The top use cases describe SSA/bootstrap analysis for interest rates, but the surrounding content steers operation toward stock/crypto trading workflows. This inconsistency can cause deceptive routing, unsafe operator assumptions, and accidental use of the skill in higher-risk financial contexts than its advertised analytical scope suggests.

Description-Behavior Mismatch

High

• Confidence: 98% confidence
• Finding: The skill’s human-facing description materially conflicts with its declared actuarial interest-rate purpose, instead advertising ZVT-based equity/crypto strategy development and backtesting. This kind of scope mismatch is dangerous because it can mislead users and orchestrators into granting the skill access, trust, or execution opportunities for trading-related tasks that were not intended or reviewed under the declared manifest purpose.

Intent-Code Divergence

High

• Confidence: 97% confidence
• Finding: The tagline and use cases actively reframe the skill as a quant trading assistant, which is a stronger signal than a minor documentation error because it shapes user behavior and task routing. In security terms, deceptive or inaccurate capability framing increases the risk of unauthorized use, unsafe delegation, and review bypass, especially if downstream systems rely on metadata and summaries to decide when to invoke the skill.

Context-Inappropriate Capability

Medium

• Confidence: 93% confidence
• Finding: The documentation advertises broad market-data and trading-related capabilities across A-share, HK, crypto, and multiple providers that are unjustified by the stated actuarial purpose. Even if no malicious code is present, this overclaiming expands the apparent operational scope of the skill and can induce users to expose sensitive financial workflows or rely on unsupported trading functions.

Description-Behavior Mismatch

High

• Confidence: 96% confidence
• Finding: The file declares trading and backtesting semantic locks for equity workflows, but the skill metadata describes actuarial interest-rate decomposition and NSS curve calibration. This mismatch is dangerous because it can smuggle in unrelated market-trading behaviors, dependencies, and execution assumptions that expand the skill's effective capability beyond its stated purpose, increasing the chance of misuse or hidden data/execution paths.

Context-Inappropriate Capability

High

• Confidence: 95% confidence
• Finding: The semantic locks explicitly encode trading signals, order execution, MACD parameters, A-share T+1 rules, transaction costs, and factor-pipeline behavior, none of which are justified for an actuarial interest-rate analytics skill. In context, this is especially concerning because it suggests hidden or unintended trading-system functionality embedded in a domain that should be limited to rate modeling and statistical inference.

Context-Inappropriate Capability

Medium

• Confidence: 92% confidence
• Finding: The preconditions require zvt installation, initialization, and recording of equity market kdata for stock_sh_600000, which is unrelated to actuarial interest-rate modeling. This unnecessary dependency broadens the attack surface, may trigger unneeded data access or environment modification, and can mislead operators into provisioning market-trading infrastructure for a non-trading task.

Description-Behavior Mismatch

Critical

• Confidence: 99% confidence
• Finding: The seed content is materially inconsistent with the declared skill purpose: it claims insurance/actuarial interest-rate analysis, but large sections instead configure ZVT-based stock trading, backtesting, market-data collection, and trading execution workflows. This kind of capability mismatch is dangerous because it can cause an agent to execute unintended financial-market actions, install unrelated tooling, or generate trading code under a misleading actuarial label, defeating user consent and review expectations.

Context-Inappropriate Capability

High

• Confidence: 96% confidence
• Finding: The file exposes stock-market trading and backtesting capabilities that are outside the stated insurance-actuarial function, including trading-oriented preconditions, routing, and execution semantics. In context, this broadens the operational surface from analysis into action-like behavior, increasing the risk of unauthorized code generation, misleading outputs, and accidental use of market/trading workflows when the user expects actuarial research tooling.

Intent-Code Divergence

High

• Confidence: 98% confidence
• Finding: The user-facing summary explicitly markets the skill as helping build A-share quant strategies with ZVT, which directly contradicts the insurance-actuarial metadata and description. This is dangerous because users and downstream agents rely on the displayed summary for trust and consent; contradictory messaging can socially engineer users into approving unrelated financial workflows or misjudge what code and data access the skill will perform.

Vague Triggers

Medium

• Confidence: 89% confidence
• Finding: The execute trigger is broad and ambiguous, combining loose intent matching with common action verbs like run, execute, fetch, or collect. In a skill that already mixes benign analytics with trading-related behavior, this can cause unintended invocation and execution in response to routine user requests.

Vague Triggers

Medium

• Confidence: 82% confidence
• Finding: The execute trigger combines broad intent matching with generic verbs like run/execute/backtest/fetch/collect, making activation conditions too permissive and ambiguous. In a skill that already mixes unrelated actuarial and trading workflows, this ambiguity increases the chance of accidental invocation of privileged or state-changing behavior when a user intended a benign analytical question.

Missing User Warnings

Medium

• Confidence: 88% confidence
• Finding: The manifest states that a .skill file will be automatically written after successful execution, but there is no equally prominent upfront warning that persistent files will be created. Silent persistence is risky because it changes the user's environment and may create durable capabilities or artifacts without informed consent, which is especially concerning in a skill already exhibiting scope confusion.

Static analysis

No suspicious patterns detected.