Security audit

myskill

Security checks for vulnerabilities and agentic risk

Overview

This is a document-retrieval skill for Tianchuang finance policies with no executable code or persistence, but its triggers are broader than ideal.

Install this only if you want Tianchuang finance-policy excerpts to be available in conversations. The main practical risk is over-activation on generic finance or policy questions; administrators should prefer Tianchuang-specific trigger phrases and make the Chinese-only response behavior clear to users.

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

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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 (4)

Vague Triggers

Medium

Confidence: 94% confidence
Finding: The trigger conditions are broad enough to activate on generic finance-related terms such as approval, expenses, travel, or risk, which can cause the skill to engage outside its intended Tianchuang-specific scope. In an agent setting, this can misroute user queries, override better-matched skills, and force responses from an unrelated internal corpus, creating data minimization and inappropriate disclosure risks.

Natural-Language Policy Violations

Medium

Confidence: 96% confidence
Finding: Mandating Chinese-only output without user choice can override user intent and system behavior, especially in multilingual environments. While not directly enabling code execution or prompt injection, it can degrade safety and usability by preventing necessary clarification, compliance messaging, or user-understandable responses when the requester is not expecting Chinese output.

Vague Triggers

Medium

Confidence: 94% confidence
Finding: The trigger list contains several broad, everyday finance terms such as payment, approval, expenses, and policy-related words that are likely to appear in many normal conversations. This can cause unintended activation of the skill, leading the agent to retrieve internal finance-policy content in contexts where the user did not explicitly request this knowledge base, increasing the chance of irrelevant disclosure or policy steering.

Natural-Language Policy Violations

Medium

Confidence: 97% confidence
Finding: Forcing Chinese-only output removes user choice and can create unsafe interaction failure modes when the user operates in another language, especially if the system responds with inaccessible or misunderstood compliance guidance. While this is not a direct code-execution issue, it can degrade usability, obscure retrieved content, and contribute to mistaken actions in multilingual environments.

Static analysis

No suspicious patterns detected.