Security audit

Ultrahuman Biodata Assistant

Security checks for vulnerabilities and agentic risk

Overview

This skill is purpose-built for Ultrahuman health metrics, but its broad triggers could access sensitive account health data when a user only asked a general sleep or wellness question.

Review this before installing if you use Ultrahuman data in a shared or sensitive environment. The skill should ideally ask before accessing the linked account for ambiguous wellness questions, confirm which email/account it will use, and retrieve only the dates and metrics needed for the specific request.

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: 94% confidence
Finding: The invocation guidance is broad enough to activate on generic sleep, recovery, or wellness questions even when the user may not be asking to access Ultrahuman data. That creates a scope-confusion risk where the assistant could unnecessarily reach for sensitive health tools and account-linked data without first confirming the user's intent.

Vague Triggers

Medium

Confidence: 89% confidence
Finding: The workflow examples like 'how did I sleep' or 'am I ready to train?' are ambiguous because they can describe general advice requests rather than requests to retrieve private device data. In context, this ambiguity increases the chance of unintentional access to sensitive health records when a safer clarification step should occur first.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The skill instructs the assistant to fetch sensitive health and metabolic data using account email and environment credentials, but it does not require explicit user notice or consent at the point of access. Because the data includes highly sensitive biometric and health information, silent use of stored credentials raises privacy and unauthorized-disclosure risks.

Static analysis

No suspicious patterns detected.