Back to skill

Security audit

Lobster Attachment Inject

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed Markdown-only guide for dynamically loading agent and skill metadata, with no bundled executable code or hidden behavior found.

Before installing, confirm you want a skill that can read agent and skill definitions and potentially maintain registry or AGENTS.md injection content. Use it with explicit routing or confirmation for loading additional agents or skills, and scope write access to the intended registry/config files.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (3)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The trigger examples are broad natural-language cues (e.g. mentioning scheduling an agent or debugging) that can cause the system to read and activate additional agent/skill content without a precise authorization or disambiguation step. In a prompt-injection setting, ambiguous activation logic increases the chance of unintended capability loading, context expansion, or execution of the wrong workflow based solely on user phrasing.

Content

No source excerpt is available for this finding.

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
81% confidence
Finding

The skill explicitly describes enumerating and reading from agents/*.md and specific skill files on demand, which exposes internal capability discovery patterns. In hostile environments, this makes it easier for an attacker to probe available agents/skills, induce loading of sensitive prompt assets, and chain prompt-injection attempts against whichever files are discovered.

Content

Scanner excerpt · SKILL.md (reported line 23)May include surrounding context.

不预注入内容,而是在需要时读取:

text
用户提到"调度Agent" → 读 agents/*.md → 选择合适的Agent
用户提到"调试" → 读 skills/systematic-debugging/SKILL.md → 激活技能

模式2: 注册表文件(轻量注入)

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
81% confidence
Finding

The skill content and examples are presented in Chinese throughout the file, and there is no indication that the user can choose another language or that the locale restriction is intentional for a region-specific use case. This can violate a language/locale policy where skills should not force a specific language without opt-in.

Content

No source excerpt is available for this finding.