Security audit

天才龙虾

Security checks for vulnerabilities and agentic risk

Overview

This is a playful Chinese-language lobster persona skill with local helper scripts and no evidence of data access, persistence, or unsafe actions.

Install this if you want a Chinese-language humorous lobster persona. Be aware it may activate for broad creative, philosophical, physics, or humor requests unless your agent's skill routing is stricter.

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
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
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Vague Triggers

Medium

Confidence: 95% confidence
Finding: The activation description includes broad triggers such as any request for a 'crustacean perspective on complex topics,' which can match many unrelated conversations and cause unintended skill activation. Over-broad activation can override the intended assistant behavior, create confusing persona shifts, and increase the chance that users receive responses in an unexpected style or language.

Natural-Language Policy Violations

Medium

Confidence: 97% confidence
Finding: The skill mandates Chinese output ('用中文') without indicating user choice, which can conflict with the user's language preference and reduce usability or transparency when the skill activates unexpectedly. In combination with the broad triggers elsewhere in this file, forced language switching becomes more disruptive because users may be moved into both an unrequested persona and an unrequested language.

Vague Triggers

Medium

Confidence: 92% confidence
Finding: The usage table contains vague trigger classes like '任何 deep、哲学、物理问题' and '用户需要幽默/反讽的回应,' which are not concrete invocation phrases and can spuriously activate the skill for a wide range of normal requests. In context, this makes the persona unusually sticky and may interfere with other skills or the base assistant by hijacking broad intellectual or stylistic queries.

Natural-Language Policy Violations

Medium

Confidence: 97% confidence
Finding: This Python skill hardcodes its description and generated analysis text in Chinese, and the function constructs only Chinese-language output. The policy scope includes language/locale violations for all file types, and there is no indication that the user can choose a language or that the Chinese-only behavior is justified as region-specific.

Static analysis

No suspicious patterns detected.