Security audit

YT Shorts Niche Research

Security checks for vulnerabilities and agentic risk

Overview

This skill is for YouTube Shorts research, but it tells the agent to run unaudited local code, retry without a clear stop limit, and create weekly unprompted Telegram delivery with persistent state.

Install only if you intentionally want this personal workflow and can inspect the external youtube-research.js file yourself. Before use, require explicit approval for every run, remove or disable the weekly Telegram/HEARTBEAT behavior, add retry and timeout limits, and avoid running mutable code from an absolute path outside the reviewed skill package.

A.I.G

Vulnerability Patterns

System PersistenceInstalls backdoors, hooks, services, or scheduled tasks that survive the run
Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
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

Findings (3)

T06 · System Persistence

Error

Location: SKILL.md:59
Finding: Autonomous Recurring Execution and Persistent Agent-State Modification
Content: View full analysis
Remediation: View remediation

T09 · Insecure Skill Coding Practices

Error

Location: SKILL.md:16
Finding: Execution of Mutable External Code Outside the Audited Skill Package
Content: View full analysis
Remediation: View remediation

T09 · Insecure Skill Coding Practices

Warning

Location: SKILL.md:27
Finding: Unbounded Immediate Retry Loop Can Cause Resource Exhaustion
Content: View full analysis
Remediation: View remediation

SkillSpector

Vulnerability Patterns

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 93% confidence
Finding: The trigger phrases are broad enough to match common research requests, increasing the chance the skill runs in situations where the user did not specifically intend this workflow. Because the skill executes local scripts and includes persistence/automation behaviors elsewhere in the file, overbroad invocation increases the risk of unintended actions.
Content: No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium

Category: Not specified by scanner
Confidence: 97% confidence
Finding: The skill adds a weekly autonomous action to send results to Abdullah on Telegram and update a heartbeat file, which goes beyond the stated on-demand research function. This creates an unsolicited outbound messaging/automation capability that can cause unintended data disclosure, persistent behavior, and actions without a fresh user request or consent.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 98% confidence
Finding: Scheduling and sending results on Telegram unprompted introduces undisclosed outbound communication to an external channel. In the context of a research skill, this is more dangerous because it is unrelated to the core task, bypasses normal user initiation, and can leak data or create persistent autonomous behavior.
Content: No source excerpt is available for this finding.

Missing User Warnings

Low

Category: Not specified by scanner
Confidence: 90% confidence
Finding: The instruction to update HEARTBEAT.md modifies a local file without warning the user that the skill will write to disk. Even though the file change is small, undisclosed state changes create persistence and can be abused to conceal activity timing or maintain operational state without user awareness.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.