Security audit

Write substack like articles

Security checks for vulnerabilities and agentic risk

Overview

This is a markdown-only writing skill whose research and drafting instructions are broadly aligned with its stated Substack newsletter purpose.

Before installing, be aware that the skill may send research queries to web search and configured MCP services, including Readwise or LlamaCloud. Avoid using it with confidential article ideas or private institutional strategy unless you are comfortable with those services receiving topic-derived queries, and review drafts for proper attribution where outside frameworks or sources informed the writing.

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
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (2)

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill instructs the agent to send topic queries to external services such as web_search, Readwise MCP, and LlamaCloud MCP without any requirement to notify the user that their prompt-derived content may be transmitted to third parties. If a user's request contains sensitive strategy, unpublished writing ideas, or personal/institutional context, the agent could disclose that information externally without informed consent.

Missing User Warnings

Low

Confidence: 82% confidence
Finding: The guidance tells the agent to extract insights, remove original author names, and rewrite material in the user's voice, but it provides no attribution or source-handling safeguards. This creates a risk of opaque paraphrasing of third-party intellectual content, which can mislead users about provenance and increase plagiarism or copyright-compliance issues.