ClawHub
Back to skill

Security audit

Reversal — Agent Input Reliability Layer

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed document and URL parsing integration, but users should treat remote mode as sending selected files or URLs to Reversal's service.

Prefer the local stdio setup for confidential files, internal URLs, regulated data, or proprietary documents. If using remote HTTP or the SDKs, assume the selected source and parsed output may be processed by Reversal's service, protect API keys, and review the external repository and service terms before installing.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (3)

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The skill documents remote HTTP and SDK usage that sends user-supplied URLs, files, and possibly extracted document content to a third-party service, but it does not clearly warn users about that data egress. In an agent setting, this can cause operators to unintentionally transmit sensitive local files, internal URLs, or proprietary content outside their trust boundary.

External Transmission

Medium
Category
Data Exfiltration
Content
"mcpServers": {
    "reversal": {
      "transport": "http",
      "url": "https://api.reversal.dev/v1/mcp",
      "headers": { "Authorization": "Bearer sk-rev-YOUR_KEY" }
    }
  }
Confidence
93% confidence
Finding
Configuring the MCP server to use the hosted HTTP endpoint means agent requests, including supplied URLs, uploaded file references, and potentially extracted content, are sent to an external service. In the context of a tool that can access local files and arbitrary sources, this creates real data-exfiltration risk if users are not clearly warned and if sensitive inputs are forwarded by default.

External Transmission

Medium
Category
Data Exfiltration
Content
```toml
[[mcp_servers]]
name = "reversal"
url  = "https://api.reversal.dev/v1/mcp"
[mcp_servers.headers]
Authorization = "Bearer sk-rev-YOUR_KEY"
```
Confidence
93% confidence
Finding
The Codex MCP configuration points the agent at a remote endpoint, enabling external transmission of agent inputs and retrieved content during normal operation. Because this skill is specifically designed to normalize URLs and files, including potentially sensitive local documents, the remote configuration materially increases the chance of unintended third-party disclosure.

Static analysis

No suspicious patterns detected.