ClawHub
Back to skill

Security audit

unisound-discharge-record

Security checks across malware telemetry and agentic risk

Overview

This skill has a legitimate medical-document purpose, but it handles sensitive patient records with network transfer and optional plaintext persistence that are not scoped clearly enough.

Review this skill before installing in any clinical or regulated environment. Only use it with de-identified records, an approved HTTPS model endpoint, an authorized app key, and a clear retention policy. Avoid --save-prepared and output paths that persist patient data unless local storage is approved and protected.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Lp3

Medium
Category
MCP Least Privilege
Confidence
93% confidence
Finding
The skill advertises and documents capabilities for reading local files, writing outputs/preprocessed data to disk, and sending medical records over the network, yet it declares no corresponding permissions. This creates a governance and user-consent gap: operators may trust the manifest while the actual workflow processes sensitive PHI and can exfiltrate it to an external/internal API and persist it locally.

Intent-Code Divergence

Medium
Confidence
97% confidence
Finding
The document claims 'no local persistence' and that data is destroyed after the call, but later exposes --output-json, --output, and --save-prepared options that write patient data or intermediate text to disk. For a medical-record skill, this contradiction is dangerous because users may rely on the privacy claim and unintentionally store sensitive health information locally, increasing leakage, retention, and compliance risk.

Missing User Warnings

High
Confidence
97% confidence
Finding
The script sends raw patient medical record content to a remote API endpoint via `call_llm`, with the endpoint configurable by `--base`. This creates a real confidentiality risk because highly sensitive PHI can be transmitted off-host or even to an attacker-controlled endpoint without any built-in consent notice, allowlist, redaction, or transport policy enforcement.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
When `--save-prepared` is used, the tool writes preprocessed patient record text to disk in plaintext. This is sensitive local persistence of medical data, which can lead to unauthorized disclosure through backups, shared directories, weak filesystem permissions, or accidental retention beyond the intended processing window.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.