unisound-literature-retrieval

• Outcome: Pass
• Latest audit: Jul 16, 2026 · 7:57 AM UTC
• Version: 1.0.2

Overview

This skill is a disclosed, user-run medical literature assistant that sends user-provided questions and passages to a named remote medical model API, but users should avoid including patient identifiers or confidential material.

Install only if you are comfortable sending the provided clinical question, constraints, and literature excerpts to the documented hivoice medical model API. Do not include patient identifiers, protected health information, unpublished research, or confidential institutional material unless your organization has approved that data flow. Treat the --appkey value as a secret.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 84% confidence
• Finding: The skill advertises execution via a Python script that reads input files, writes output files, and sends data to a remote API, but the manifest does not declare any permissions or clearly expose those capabilities to reviewers/users. This creates a transparency and governance gap: sensitive clinical questions and literature excerpts may be processed and transmitted off-box without an explicit permission model, increasing the risk of unauthorized data handling.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The skill states that user-provided clinical questions and optional literature passages are sent to an internal medical model API, but it does not present an explicit user warning or consent notice near the description or usage flow. In a medical context, these inputs can contain sensitive patient-related details, proprietary research material, or unpublished excerpts, so silent transmission to a remote service creates a meaningful privacy and compliance risk.

Missing User Warnings

Medium

• Confidence: 98% confidence
• Finding: The script sends clinical questions, constraints, and user-supplied literature excerpts to an external API, and those inputs may contain sensitive clinical or research data. In a medical context, undisclosed transmission to a remote third party can create confidentiality, compliance, and data-governance risks, especially if patient-identifying information is included.

Missing User Warnings

Low

• Confidence: 93% confidence
• Finding: Accepting the API credential via a command-line argument exposes it to local process inspection, shell history, job control logs, and audit tooling on many systems. While this is a common convenience pattern, it is unsafe for secrets because other local users or monitoring systems may be able to recover the token.

VirusTotal

61/61 vendors flagged this skill as clean.

Static analysis

No suspicious patterns detected.