Security audit

research-monitor

Security checks for vulnerabilities and agentic risk

Overview

This Korean research skill is coherent, but it describes automatic ongoing monitoring and Telegram alerts without enough user control or scoping.

Install only if you want a Korean-language monitoring workflow that may run on schedules and send results to Telegram. Before enabling it, confirm the Telegram recipient, monitored keywords and sites, alert frequency, retention expectations, and how to pause or disable the scheduled and real-time alerts.

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 (4)

Vague Triggers

Medium

Confidence: 95% confidence
Finding: The trigger phrases are common everyday requests such as asking for news summaries or trend analysis, which creates a realistic risk of accidental invocation outside a clearly consented workflow. Because the skill can initiate monitoring behavior and downstream notifications, ambiguous activation increases the chance of unintended data collection or actions.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The skill explicitly schedules automatic industry news crawling and Telegram transmission, but does not warn users that collected results will be sent to an external messaging platform. This is dangerous because users may not understand that monitoring output leaves the local environment and may be continuously delivered to third-party services.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: Real-time keyword alerts imply continuous monitoring and external notification behavior, but the skill does not disclose retention, frequency, or where notifications are sent. In a monitoring context, missing disclosure and consent can lead to persistent outbound data flows that users did not knowingly authorize.

Natural-Language Policy Violations

Low

Confidence: 67% confidence
Finding: All headings, descriptions, and trigger phrases are presented only in Korean, and no alternative language option or opt-in is documented. Under the language/locale policy, forcing a single language without user choice can be a natural-language policy concern unless clearly justified.

Static analysis

No suspicious patterns detected.