ClawHub
Back to skill

Security audit

long-task-handler-v2

Security checks for vulnerabilities and agentic risk

Overview

This Markdown-only skill manages long-running tasks and progress updates; its scope is broad, but the behavior is disclosed and purpose-aligned.

Install only if you want the agent to proactively handle likely long-running work in the background and send periodic updates. Review the broad activation rules if you prefer explicit confirmation before background execution or sub-agent use.

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
  • 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 (2)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill description and metadata define a very broad activation scope for any task expected to exceed 60 seconds, including compilation, deployment, training, data processing, crawling, writing, and external API usage. In an agent system, such broad scope can cause the skill to activate in many ordinary workflows without clear exclusions or precedence rules, leading to unintended background execution, excessive status messaging, or inappropriate delegation behavior.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger list includes ambiguous categories such as '预计耗时 > 60 秒', common phrases like '别等我', and broad command/data heuristics such as file count > 100 or generic slow commands. These conditions are easy to match unintentionally, which can alter task handling semantics, mask normal synchronous execution behind background behavior, and create unpredictable agent behavior across benign user requests.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.