Back to skill

Security audit

Danke Blog Writer

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed Chinese personal blog-writing skill with optional user-confirmed publishing and only limited local helper scripts.

Install only if you want a Chinese personal blog-writing assistant for this specific style/profile. Before using the Halo publish step, review and explicitly approve the destination, title, content, and tags.

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
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (13)

Tp4

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding

描述把该技能定位为一个端到端的博客写作助手,但提供的代码只是其中一个子模块:文本人性化处理脚本。它没有实现选题策划、标题生成、草稿撰写、发布流程,也没有体现完整工作流编排。虽然“去 AI 味”确实与声明中的一部分相符,但实际代码的主要行为明显比声明窄得多,且加入了面向多个内容平台的格式化适配,更像通用润色工具而非专属博客全流程助手。因此应判定为描述与行为不匹配。

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The declared description presents a content creation assistant for end-to-end blog writing and publishing. The supplied code does not generate topics, titles, drafts, style edits, or publishing actions. Instead, it manages a local examples library by enumerating markdown files, checking whether the count exceeds a limit, recommending old files for removal, and optionally deleting them. This is a materially different primary purpose and introduces an undeclared file-deletion capability.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The README advertises one-click publication to a live external blog endpoint without any explicit warning, confirmation requirement, or mention of publishing side effects. In agent contexts, that can normalize high-impact actions and increase the risk of accidental or unauthorized publication of drafts, sensitive content, or hallucinated material.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The usage section activates on very general phrases like “帮我写篇关于 xxx 的博客” and “给这个草稿起几个标题,” without defining any constraints, exclusions, or narrower context for when this skill should be invoked. These phrases are common in everyday assistant interactions and may cause unintended activation collisions with other generic writing tasks.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The trigger conditions are broad enough to capture generic writing-assistance requests, which can cause the skill to activate in situations the user did not explicitly intend. In a skill ecosystem, overbroad routing increases the chance of unexpected behavior, including accidental use of downstream tooling or applying a strong persona/style transformation without clear consent.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The skill includes publishing to a live Halo blog endpoint and references a specific profile, but it does not require a prominent user-facing confirmation or warn that this is an external write action. That creates risk of accidental publication, unintended content changes, or posting sensitive draft material to a production site.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The entire reference file is written in Chinese and presents all examples, labels, and usage guidance exclusively in Chinese. Under the policy for natural-language violations, forcing a specific language without explicit user choice or documented locale limitation is a reportable issue.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
82% confidence
Finding

The statement that different platforms require different writing styles positions the document as a normative instruction set for how content should be written. Across the file, the guidance fixes language/tone conventions per platform, but does not mention user opt-in, preference selection, or a justified locale/policy exception.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This code is built around hard-coded Chinese vocabulary substitutions and later adds Chinese-language prefixes such as '【分享】' and '个人经验:'. Because the skill does not offer any language or locale selection beyond platform choice, it effectively imposes a specific language/locale behavior without user opt-in.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
81% confidence
Finding

The README describes the skill as a dedicated assistant for a specific person in Chinese and does not indicate that users may choose another language or locale. This can be a natural-language policy concern when a skill implicitly constrains language behavior without opt-in or documented justification.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

The manifest description is written as a dedicated assistant for a specific person and presents the skill in Chinese only, with no indication that users can choose another language or locale. This can violate the language/locale policy when the skill appears to impose a specific language without explicit opt-in or justification.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

This markdown file presents all instructions and examples exclusively in Chinese, and there is no indication that the skill is region-specific or that users can opt into this language. Under the policy, forcing a specific language without user choice can be a natural-language policy violation.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

add_personal_voice is documented at L087-L096 as increasing first-person voice, which aligns with the manifest's promise of personal-style writing. However, humanize at L099-L117 performs filler removal, word replacement, pattern fixes, dash simplification, and platform adaptation only; it never calls add_personal_voice, so the documented intent is not actually implemented.

Content

No source excerpt is available for this finding.