Back to skill

Security audit

Claude Code Mastery

Security checks for vulnerabilities and agentic risk

Overview

This is a mostly informational Claude Code guide with prompt templates, and I did not find executable code, hidden persistence, or automatic data access.

Before using the MCP portions, connect only services you intend Claude Code to access, prefer read-only or least-privilege credentials, and test against non-production data first. The included files are guide and template content, not active code.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The provided manifest context describes a skill whose purpose is teaching or guiding users through Claude Code CLI workflows. This file instead declares a separate skill named 'Security Audit' that checks code for vulnerabilities, which is a materially different function from the stated skill purpose.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The skill explicitly promotes MCP servers for database access, API testing, GitHub/GitLab integration, and file system operations without any warning about credentials, sensitive data exposure, destructive actions, or least-privilege setup. In a guide aimed at broad audiences, this omission can normalize connecting high-impact tools to an agent without adequate safeguards, increasing the chance of accidental data leakage or harmful system actions.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This manifest uses broad natural-language phrases like "Review changed files in current session" and "Review the changes I made in this session" without defining clear trigger constraints or exclusions. Those phrases overlap with common everyday requests for assistance and do not specify narrow activation conditions, increasing the risk of unintended invocation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
71% confidence
Finding

The listed key commands include "/番茄", which introduces a specific language/locale element in the skill description without clarifying whether language choice is optional or user-configurable. Under the stated policy, forcing or implying a specific language without opt-in can be a natural-language policy concern.

Content

No source excerpt is available for this finding.