ClawHub
Back to skill

Security audit

technical-article-writer

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed Markdown-only technical writing skill with some broad writing and marketing-adjacent workflow steps, but no hidden execution, persistence, credential use, or destructive behavior.

Install this if you want a guided technical article workflow. Be aware it may ask detailed intake questions, use optional web lookup for Midjourney prompt syntax, and delegate hooks, CTA, or style cleanup to other installed skills, so review those helper skills separately if they are present.

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
  • Memory PoisoningPersistent Context Injection, Context Window Stuffing, Memory Manipulation
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (5)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The activation text says to use the skill whenever the user asks to write a blog post, technical article, or 'any long-form technical content,' and also when the user says phrases like 'help me draft an article' or even 'I want to write about X' without specifying format. These triggers are broad and ambiguous enough to overlap with many ordinary writing requests, increasing the risk of unintended invocation.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The manifest explicitly says not to use this skill for newsletter strategy, but Phase 1 includes 'Newsletter subscription / audience growth' as a primary objective and Phase 5b passes that objective into a CTA skill to design conversion mechanisms. That goes beyond article drafting into promotional/newsletter strategy.

Content

No source excerpt is available for this finding.

Context Window Stuffing

Medium
Category
Memory Poisoning
Confidence
80% confidence
Finding

Skill attempts to fill the context window with filler content, displacing legitimate instructions and safety constraints. This can degrade agent performance or bypass safety boundaries.

Content

Scanner excerpt · SKILL.md (reported line 69)May include surrounding context.

md
- If you can't infer a thesis, ask the user: "What's the one thing you want the reader to remember?" If they can't answer, help them find it through questions about what surprised them, what they'd tell a colleague, or what they wish they'd known earlier.
- If the content type is ambiguous (could be a tutorial or an explainer), ask which experience the reader should have: following along hands-on, or building a mental model.

Only proceed to Phase 2 once you have enough clarity on topic, audience, content type, and thesis to write a coherent outline. It's cheaper to ask one question now than to rewrite 2000 words later.

**Idea quality filters.** Apply these before investing in a draft:

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The manifest says the skill should not be used for platform-specific optimization, yet Phase 6 instructs the agent to generate prompts using the latest Midjourney model conventions and current syntax parameters. That is a concrete platform-specific content optimization behavior beyond general technical article writing.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Low
Category
Not specified by scanner
Confidence
78% confidence
Finding

The manifest says not to use the skill for ghostwriting voice matching, yet the Humanize phase explicitly rewrites the article to remove AI patterns and make it sound more natural. While not explicit persona mimicry, this is adjacent style-transformation behavior not clearly covered by the stated article-writing scope and close to the excluded use case.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.