ClawHub
Back to skill

Security audit

Publish & Share HTML games

Security checks for vulnerabilities and agentic risk

Overview

PaperBox does what it says by publishing HTML projects to a sharing service, but its instructions are broad enough that an agent could upload a project publicly without clear, explicit confirmation.

Install only if you want agents to publish completed HTML projects to PaperBox. Before using it, confirm the project is meant to be public or shareable, and do not upload secrets, tokens, private customer data, proprietary code, or other sensitive content.

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 (3)

Vague Triggers

Medium
Confidence
96% confidence
Finding
The skill advertises very broad trigger phrases such as 'share this', 'publish my app', and 'get a link', which can cause the agent to invoke the skill in contexts where the user did not explicitly consent to uploading content to a third-party service. Because this skill performs a network publish action, overbroad invocation increases the risk of unintended disclosure of generated or user-provided content.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The workflow instructs the agent to upload full HTML content to a remote API and return a shareable URL, but it does not require informing the user that the content is being transmitted off-platform and made publicly accessible. This creates a meaningful privacy and data-governance risk because users may not realize the upload is external and the result is a public link.

External Transmission

Medium
Category
Data Exfiltration
Content
### Request

```bash
curl -X POST https://paperbox-beta.vercel.app/api/games \
  -H "Content-Type: application/json" \
  -H "X-API-Key: <from openclaw.json skills.entries.paperbox.apiKey>" \
  -d '{
Confidence
91% confidence
Finding
This skill performs external transmission of project content to a third-party endpoint and includes use of an API key from local configuration. While external upload is the intended function, it is still a real security concern because it can exfiltrate sensitive content if invoked on private material, and the instruction set lacks safeguards around secret scanning, consent, and data classification.

Static analysis

No suspicious patterns detected.