hit-content-writer

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:42 PM UTC
• Version:
  1.0.0

Overview

This is a coherent markdown-only content rewriting skill, but users should be careful with stored text, embedding API keys, and third-party source content.

Install only if you trust the external jl-vector-store tool and embedding provider. Use a scoped API key, avoid storing personal or confidential text, verify you have rights to extract or rewrite third-party content, and check how to delete or manage the ChromaDB data before relying on it.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (10)

Vague Triggers

Medium

• Confidence: 94% confidence
• Finding: The extraction operation can be triggered by broad natural-language phrases like '提取文案' or '获取文案', which are common expressions a user might say conversationally. In an agent environment, this creates a prompt-trigger risk where normal discussion may unintentionally invoke content extraction against a supplied link or shared text.

Vague Triggers

Medium

• Confidence: 95% confidence
• Finding: The content-analysis triggers are generic phrases such as '分析内容' and '智能分析', which overlap with ordinary user requests and make intent boundaries unclear. This increases the chance that unrelated text in a conversation is treated as executable skill input, causing unintended processing of user-provided or sensitive content.

Vague Triggers

Medium

• Confidence: 96% confidence
• Finding: The storage operation uses broad trigger phrases like '存储内容' and '保存内容', which can easily match ordinary requests. Because this action persists data, accidental activation is more serious: users may unintentionally store proprietary, personal, or regulated content into a local or external vector store.

Vague Triggers

Medium

• Confidence: 93% confidence
• Finding: The query triggers overlap with common conversational search phrases such as '查询内容' or '搜索内容', so users may invoke retrieval without realizing they are querying a persistent content database. This can expose previously stored material in contexts where the user intended only a general question, not database retrieval.

Vague Triggers

Medium

• Confidence: 92% confidence
• Finding: The rewrite operation is enabled by broad phrases like '仿写内容' and '智能仿写', which are semantically close to ordinary content-assistance requests. In an agent setting this can cause unplanned generation based on referenced materials, including possible misuse of copyrighted, sensitive, or policy-restricted source text.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: The README advertises storage and vector-database querying but does not clearly warn users that submitted content may be persisted and sent for embedding to an external service. This omission undermines informed consent and can lead to accidental disclosure or long-term retention of sensitive text.

Missing User Warnings

Medium

• Confidence: 89% confidence
• Finding: The copy-extraction feature describes scraping or extracting text from third-party platform links without any explicit caution about privacy, terms-of-service, copyright, or compliance implications. Users may unknowingly process protected or personal content, creating legal and data-handling risk.

Vague Triggers

Medium

• Confidence: 94% confidence
• Finding: The trigger list uses very broad, generic phrases such as '分析内容', '查询内容', and '仿写内容', which can easily overlap with normal user requests and cause unintended skill activation. Because this skill can invoke other skills and a storage/query tool, accidental activation could lead to unnecessary data retrieval, content processing, or persistence without the user clearly intending to use this specific skill.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The document explicitly promotes imitation and rewrite workflows based on reference 'viral' content, including preserving structure and style and targeting similarity ranges, but it provides no safeguards around copyright, plagiarism, or substantial similarity. In a content-generation skill, this can lead users to create infringing or deceptively derivative material at scale, increasing legal and trust risks.

Missing User Warnings

Medium

• Confidence: 88% confidence
• Finding: The skill describes collecting user feedback, tracking content performance, and building knowledge bases from usage outcomes without disclosing data handling limits, consent expectations, retention, or privacy protections. Even though this is only a strategy document, it normalizes data collection practices that could expose user-generated content, behavioral data, or platform metrics without appropriate controls.

VirusTotal

66/66 vendors flagged this skill as clean.