Back to skill

Security audit

executing-plans

Security checks for vulnerabilities and agentic risk

Overview

This skill is not clearly malicious, but it asks for command execution and credential/file handling with broad automatic-trigger language and weak user-control boundaries.

Install only if you intend to use it for bounded, user-reviewed plan execution. Before running it, require explicit approval for each command, avoid sharing broad API keys or sensitive files, and do not follow generic administrator-permission advice unless a specific task truly requires it.

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 (5)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The top-level description combines broad automation, development, data analysis, and orchestration use cases without clearly limiting when the skill should activate. In a skill with exec access, this ambiguity increases the chance of overreach, where the agent may apply the skill to tasks involving sensitive files, commands, or credentials without sufficiently explicit user intent.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

Line L043 states the skill is '基于github来源验证' to ensure data accuracy and traceability, which is a concrete behavioral claim. Elsewhere, the file exposes only generic tooling and broad operational guidance, with no actual GitHub-specific logic or verification steps, so the documentation actively overstates what the skill does.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

This section instructs users to configure API keys, prepare files, and execute commands, but it does not prominently warn that these actions may modify the local system, access sensitive data, or incur external side effects. Because the skill advertises both file handling and command execution via exec, the lack of explicit safety gating materially raises the risk of destructive or privacy-impacting actions being taken without informed consent.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The trigger condition '当用户需要执行相关操作时自动触发' is overly broad for a skill that exposes read and exec capabilities. Ambiguous activation boundaries can cause the agent to invoke the skill in contexts the user did not explicitly authorize, increasing the risk of unintended command execution, file access, or workflow changes.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
81% confidence
Finding

The SKILL.md presents a concrete JSON response contract with fields like success, data.result, and error. However, this file contains no executable implementation or prompt logic enforcing that output structure, so the documentation asserts behavior not actually present in the skill definition.

Content

No source excerpt is available for this finding.