Back to skill

Security audit

马斯克自适应推理

Security checks for vulnerabilities and agentic risk

Overview

This skill only changes the agent's internal response-depth heuristic and is broadly scoped, but that behavior is disclosed and no code, network, credential, file, or persistence capability is present.

Install this only if you want the agent to evaluate every prompt for complexity and sometimes add reasoning-mode icons at the end of responses. It appears low-risk from a security perspective because it ships no executable code and requests no data access, but its always-on behavior may be distracting or undesirable for users who prefer predictable, minimal responses.

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
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • 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 (3)

Vague Triggers

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill metadata and description state that it runs on every user message, making its trigger scope global rather than narrowly scoped. In an agent system, always-on preprocessing can override user intent, consume resources on trivial prompts, and create unsafe interaction patterns by silently changing reasoning behavior for unrelated tasks.

Content

No source excerpt is available for this finding.

Vague Triggers

High
Category
Not specified by scanner
Confidence
94% confidence
Finding

The automatic activation section instructs the agent to enable reasoning silently and 'without asking,' based on subjective scoring thresholds. This broad autonomous behavior is dangerous because it can cause hidden policy changes across conversations, bypass transparency, and encourage the agent to reveal internal-mode indicators or alter outputs in ways the user did not request.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The manifest description is written entirely in Chinese and presents the skill behavior as a default without any indication that language is configurable or user-selected. Under the stated policy, forcing a specific language or locale without opt-in can be a natural-language policy violation.

Content

No source excerpt is available for this finding.