Security audit

Bayesian reasoning

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed Bayesian reasoning helper with a local calculator script and no evidence of hidden network, credential, privilege, or destructive behavior.

Safe to install for Bayesian probability analysis. Be aware that saved analyses are written as local JSON files in the current working directory and may include the hypothesis and evidence text you entered, so avoid saving sensitive material unless that is acceptable.

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

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
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

Findings (5)

Tp4

High

Category: MCP Tool Poisoning
Confidence: 91% confidence
Finding: The code substantially matches the analytical Bayesian functionality in the description: it computes posterior probabilities via Bayes' theorem, calculates Bayes factors, performs sensitivity analysis over parameter ranges, and includes ASCII-style bar output. However, the declared purpose emphasizes guiding users through Bayesian reasoning via multi-turn dialogue and interactive learning, while the supplied code is only a non-conversational command-line calculation engine. This is a material mismatch in primary interaction model/capability, though not a security/resource mismatch.
Content: No source excerpt is available for this finding.

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 95% confidence
Finding: The manifest description says to use the skill when the user wants to 'reason about probabilities,' 'update beliefs with evidence,' or 'learn Bayesian thinking interactively.' These triggers are broad natural-language intents rather than a narrow invocation scope, and the file provides no explicit exclusions or negative examples to distinguish when this skill should not activate.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 98% confidence
Finding: The skill body is written as operational instructions in Chinese and provides no user choice or opt-in for language or locale. That creates a natural-language policy issue because it effectively constrains interaction language without documenting a justified region-specific scope.
Content: No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium

Category: Not specified by scanner
Confidence: 93% confidence
Finding: The skill instructs saving and reloading analysis results from local JSON files in the current working directory, which introduces unnecessary local data persistence for an otherwise interactive reasoning task. If users analyze sensitive hypotheses or evidence, those contents may be written to disk and later exposed to other processes, users, or sessions, especially with predictable filenames based on hypothesis keywords.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 93% confidence
Finding: This markdown file contains user-facing documentation primarily in Chinese, including the overview, field descriptions, and operational notes. Under the stated policy, forcing a specific language without user opt-in is a natural-language policy violation unless the locale restriction is clearly justified, which is not provided here.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.