Security audit

transcripcion-youtube-formula100k

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a coherent YouTube-to-script workflow, but it includes optional live Instagram publishing without clear per-post approval safeguards.

Install only if you are comfortable with a workflow that may save generated scripts locally and send them through Telegram. Do not connect or use the optional Instagram MCP unless you want the agent to have live posting capability, and require explicit approval for every post, including the target account and exact media/caption.

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
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
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

Findings (8)

Vague Triggers

High

Confidence: 97% confidence
Finding: La condición de activación es excesivamente amplia (`SIEMPRE`, `cualquier variación`) y puede disparar la skill ante casi cualquier URL de YouTube relacionada con creación de contenido. Esto aumenta el riesgo de ejecución no deseada, uso de conectores externos y procesamiento de contenido que el usuario quizá no quiso transcribir o reutilizar todavía.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: La skill indica que los resultados se guardan en `guiones/` y se envían por Telegram sin exigir un aviso o consentimiento previo del usuario. Si la transcripción o los guiones contienen material sensible, privado o protegido, esta salida y transmisión automática puede provocar exposición de información a almacenamiento local y a un canal externo.

Natural-Language Policy Violations

Medium

Confidence: 88% confidence
Finding: La instrucción de L049 dice que si el video está en inglés, se extraiga en inglés y luego se adapte al español en los guiones. Esto fuerza un idioma de salida específico sin indicar que el usuario pueda optar por otro idioma o variante lingüística.

Natural-Language Policy Violations

Medium

Confidence: 86% confidence
Finding: En L184 se instruye que, si el video está en otro idioma, se traduzca y adapte "al español del usuario". Aunque intenta personalizar la variante, sigue asumiendo español como idioma objetivo por defecto, sin ofrecer una elección explícita de idioma.

Context-Inappropriate Capability

Medium

Confidence: 93% confidence
Finding: The manifest describes a skill focused on obtaining a YouTube transcript and turning it into batches of scripts. Lines L209-L210 add a capability to publish reels to Instagram, which goes beyond transcription and script generation and is not justified by the declared purpose.

Context-Inappropriate Capability

Medium

Confidence: 94% confidence
Finding: The skill is scoped around YouTube transcription and script generation, but this document introduces Instagram publishing actions that can affect an external live account. Expanding capability scope in supporting docs can cause an agent to invoke unrelated high-impact tools without the user clearly intending account posting, increasing the risk of unauthorized or surprising actions.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The markdown describes direct use of `publish_reel` and `publish_photo` without any warning, approval checkpoint, or mention of irreversible effects. In an agent setting, documenting live publishing as a normal next step can normalize execution of account-affecting actions and lead to accidental posting to a real Instagram Business/Creator account.

Natural-Language Policy Violations

Low

Confidence: 84% confidence
Finding: La lista de verificación exige 'Lenguaje simple (nivel sexto grado)' como criterio obligatorio. Esto fija una política de estilo/registro lingüístico sin indicar que sea opcional, configurable o justificada por un contexto regional o de cumplimiento.

Static analysis

No suspicious patterns detected.