Back to skill

Security audit

HTML一键托管

Security checks for vulnerabilities and agentic risk

Overview

This is a documentation-style skill for using WebToWork features, with no local executable code or hidden persistence, though users should be careful with read-write platform keys and publishing actions.

Install only if you use WebToWork. Prefer read-only MCP keys for lookups, use read-write keys only for a narrow task, and explicitly confirm before publishing, deleting, changing prices, configuring webhooks, or submitting HTML that may contain private data.

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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · .clawhubignore (reported line 7)May include surrounding context.

text
*.jpeg
*.gif
*.ico
.env
.DS_Store

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The skill invites activation through very broad natural-language requests covering many sensitive capabilities, including cloning websites, publishing content, packaging apps, and configuring integrations. That increases the chance an AI agent will over-trigger this skill in ambiguous contexts and take actions beyond what the user clearly intended, especially when connected to write-capable MCP tools.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The document tells users to provide a read-write MCP key so the AI can create, modify, or delete resources, but it does not pair that instruction with a strong warning about destructive changes, unintended publication, or the need for confirmation. In an agent setting, this can normalize broad delegated authority and lead to unauthorized or accidental state changes if the model misinterprets a request.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The mini-program AI flow says that sending HTML to the AI can publish a page, including anonymous temporary publication, without clearly warning that the content may become publicly accessible and could expose embedded secrets, personal data, or unreviewed material. This creates a meaningful risk of accidental data exposure or unintended live deployment through a casual chat interaction.

Content

No source excerpt is available for this finding.