ClawHub
Back to skill

Security audit

HTML一键托管

Security checks for vulnerabilities and agentic risk

Overview

This is a documentation-only WebToWork reference skill with expected account-key guidance and no hidden code, persistence, or local execution behavior.

Install only if you use WebToWork. Use read-only MCP keys for viewing data, give read-write keys only to trusted clients when you want specific changes, confirm targets and prices before publishing or modifying resources, and rotate or revoke wk_ keys if exposed.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Vague Triggers

Medium
Confidence
94% confidence
Finding
The skill encourages invocation through very broad natural-language prompts spanning many high-impact operations, including publishing, cloning, translation, webhook setup, and monetization changes, without clear guardrails about when the agent should only advise versus when it may act. In a tool-using AI client, this ambiguity can cause over-triggering or unintended execution of sensitive actions from casual user phrasing.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The document tells users to create MCP keys and notes that read-write keys can create, modify, and delete resources, but it does not prominently treat the keys as secrets or warn that sharing them with an AI client grants broad account access. Because the same `wk_` key also works against the REST API, exposure or careless use could lead to account takeover of platform actions, destructive changes, data exfiltration, or abuse within rate limits.

Static analysis

No suspicious patterns detected.