Security audit

scientify

Security checks for vulnerabilities and agentic risk

Overview

The skill is a small research-assistant skill with a disclosed local environment check, though its marker file stores more local identity details than the documentation spells out.

Before installing, be aware that running the Quick Start command creates a file in your home directory containing your hostname, username, and platform. This is not evidence of exfiltration or destructive behavior, but users in shared or sensitive environments may prefer to inspect or skip the script.

SkillSpector

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
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
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration

Findings (4)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 91% confidence
Finding: The skill advertises and instructs execution of a local Python script that checks the environment and writes a local marker file, but the manifest declares no permissions for environment access or file writes. This mismatch weakens transparency and consent: users or orchestrators may treat the skill as low-privilege while it actually performs local inspection and filesystem modification.

Description-Behavior Mismatch

Medium

Confidence: 85% confidence
Finding: The script persists host, username, and platform details to a file in the user's home directory even though an environment check could be performed without leaving behind identifying metadata. This creates unnecessary local data collection and retention, which can expose user or system information to other local processes, backups, or later exfiltration if the workspace is compromised.

Context-Inappropriate Capability

Medium

Confidence: 90% confidence
Finding: The utility collects hostname and username metadata and saves it locally without a clear research-assistant need for that information. Even though the data is not highly sensitive by itself, it supports fingerprinting of the user environment and broadens the amount of information available if the skill or host is later inspected or abused.

Missing User Warnings

Low

Confidence: 80% confidence
Finding: The script writes a summary file before providing meaningful notice, so users may be unaware that local state and identifying metadata are being persisted. This is a transparency and consent problem that reduces user control and can make otherwise minor data collection more risky in shared or sensitive environments.

Static analysis

No suspicious patterns detected.