Security audit

generador-artifact-educativo-formula100k

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed Spanish educational HTML-generator workflow, with some external-delivery and CDN dependency considerations but no artifact-backed malicious behavior.

Install only if you are comfortable using this in a Telegram-based workflow and sending the generated educational HTML back through Telegram. Do not provide confidential student data, private business material, or copyrighted course text unless you are authorized to include it in the generated artifact. Expect the produced HTML to require internet access for CDN libraries unless the template is changed to bundle dependencies locally.

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

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (10)

Ae1

High

Category: analysis-evasion
Content: - `patrones-visuales.md` — Librería de componentes visuales reutilizables (mockups, accordion, decision matrix, etc.)
Confidence: 100% confidence
Finding: Referenced artifact was not completely inspected

Ae1

High

Category: analysis-evasion
Content: - `patrones-visuales.md` — Librería de componentes visuales reutilizables (mockups, accordion, decision matrix, etc.)
Confidence: 100% confidence
Finding: Referenced artifact was not completely inspected

Ae1

High

Category: analysis-evasion
Content: - `patrones-visuales.md` — Librería de componentes visuales reutilizables (mockups, accordion, decision matrix, etc.)
Confidence: 100% confidence
Finding: Referenced artifact was not completely inspected

Description-Behavior Mismatch

High

Confidence: 96% confidence
Finding: The manifest says the skill saves the generated file to a specific directory and sends it via Telegram, but this file contains only client-side UI code for the artifact itself. There is no code here for filesystem output or Telegram delivery, so the documented behavior is not reflected in the actual implementation shown.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The skill explicitly says generated artifacts should be sent by Telegram, but it does not require confirming that the generated file is safe to transmit or warning the user that source-derived content may leave the local environment. If the source material contains proprietary course content, student data, or sensitive business information, the agent could exfiltrate that content to a third-party messaging platform without informed consent.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The workflow repeats the instruction to send the completed HTML by Telegram as a document after validating it, again without any privacy notice, destination confirmation, or data-classification step. In context, the skill processes user-supplied educational source files, so the generated artifact can reproduce sensitive or proprietary content and be transmitted externally by default.

Natural-Language Policy Violations

Medium

Confidence: 97% confidence
Finding: The skill mandates a specific language variant ('español neutro') and forbids another valid regional form ('voseo argentino') as a strict rule. This is a natural-language policy issue because it forces a locale/style choice without indicating user opt-in or a necessary region-specific compliance reason.

Natural-Language Policy Violations

Medium

Confidence: 95% confidence
Finding: Line L498 states that texts that teach something 'van escritos de verdad en español', which is a natural-language requirement forcing Spanish. This is a locale/language policy concern because the file does not offer opt-in, alternatives, or a documented region-specific reason for restricting content to Spanish.

Natural-Language Policy Violations

Medium

Confidence: 94% confidence
Finding: The HTML root element sets `lang="es"`, and the entire UI text is written in Spanish, which establishes a fixed language choice for the skill template. The file does not indicate that users may opt into another language or that the Spanish-only behavior is required for a region-specific purpose.

Description-Behavior Mismatch

Medium

Confidence: 95% confidence
Finding: The template is not actually self-contained because it loads React, ReactDOM, Babel, and Tailwind from third-party CDNs at runtime. This creates supply-chain and availability risk: if a CDN is compromised, blocked, or unavailable, the artifact may execute untrusted code or fail to function despite claims that it works offline by double-clicking.

Static analysis

No suspicious patterns detected.