ClawHub
Back to skill

Security audit

invoice-auto-forward

Security checks across malware telemetry and agentic risk

Overview

This skill does sensitive mailbox and invoice forwarding work, but its behavior is disclosed, purpose-aligned, and user-configured.

Use this only on mailboxes and invoices you are authorized to process. Verify the forwarding and CC recipients, run scan before run, consider setting link_domains instead of allowing all invoice links, avoid --no-verify for normal setup, and install dependencies in a virtual environment.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
95% confidence
Finding
The skill clearly performs sensitive operations—reading mail via IMAP, sending mail via SMTP, writing config/secrets/state files, invoking pip/shell commands, and fetching remote links—yet no explicit permissions are declared. This creates a transparency and consent gap: users or orchestrators may invoke a skill with credential, filesystem, network, and package-install capabilities without an appropriate permission boundary or review signal.

Context-Inappropriate Capability

Medium
Confidence
89% confidence
Finding
The skill can install packages into the current Python environment during a health check, which exceeds the narrow invoice-forwarding purpose and mutates the host runtime. In an agent or shared environment, this increases supply-chain risk and can alter other tools or workflows unexpectedly.

Missing User Warnings

Medium
Confidence
90% confidence
Finding
The README promotes unattended mailbox scanning, invoice extraction, link downloading, and automatic forwarding, but it does not clearly warn users about the privacy and data-handling implications of processing invoices and email contents. Because invoices often contain sensitive personal, financial, and corporate information, deploying this skill without explicit consent, retention, and recipient-scope warnings increases the risk of accidental data disclosure or policy violations.

Tool Parameter Abuse

High
Category
Tool Misuse
Content
sp.add_argument("--link-domains", nargs="*",
                    help="仅下载这些域名后缀的链接（空=全部），如 myqcloud.com tencent.com")
    sp.add_argument("--link-timeout", type=int, help="链接下载超时秒")
    sp.add_argument("--no-verify", action="store_true", help="跳过 IMAP/SMTP 连通性验证")

    sp = sub.add_parser("check", parents=[parent])
    sp.add_argument("--install-deps", action="store_true",
Confidence
78% confidence
Finding
--no-verify

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.