Back to skill

Security audit

danjiegun

Security checks for vulnerabilities and agentic risk

Overview

This is a Chinese-language personal brand positioning skill with branded methodology content, but it does not execute code, access data, or request unusual authority.

Install this if you want guidance in this specific Chinese-language branded positioning framework. Review or override the opening, closing, and author-attribution wording if you need neutral consulting output or another language.

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
Findings (1)

T01 · Skill Instruction Hijacking

Warning
Location
references/提问清单.md:3
Finding

Mandatory Promotional Content Alters Agent Responses

Content
View full analysis

Vulnerability Details

File Location: references/Question_Checklist.md (original filename: the checklist reference file), lines 3-9 and 151-157; references/Three_Section_Staff_Positioning_Template.md (original filename: the positioning template reference file), line 131; SKILL.md, lines 1-5 and 221-225
Vulnerability Type: T01: Skill Instruction Hijacking
Risk Level: Medium

Evidence

The following English translations preserve the meaning of the relevant source excerpts.

references/Question_Checklist.md, lines 3-9:

markdown
> Complete question list for a 30-minute positioning conversation
> Ask the questions in order, record the answers, and finally output the positioning report

---

## Opening Statement

"Today I will use Captain Lao Wang's 'Three-Section Staff Positioning Method' to help you organize your personal-brand positioning. This method divides positioning into three parts..."

references/Question_Checklist.md, lines 151-157:

markdown
## Closing Statement

"This is your Three-Section Staff positioning. Remember: the work must be strong, the image must be stable, and the spirit must be genuine. Only when all three sections are connected can they exert force. It is recommended that you organize today's content onto one page, place it somewhere visible, and review it before creating content to ensure that the three sections remain connected."

---

*Question checklist completed, with 58 questions in total. A positioning conversation is expected to take 30-40 minutes.*

references/Three_Section_Staff_Positioning_Template.md, line 131:

markdown
*Three-Section Staff Positioning Method—only when all three sections are connected can they exert force.*

SKILL.md, lines 1-5 and 221-225:

markdown
# Three-Section Staff Positioning Method

> Captain Lao Wang's original personal-brand positioning methodology
> A practical summary of
...[truncated 2992 chars]
Remediation
View remediation

Remediation Suggestions

  1. Remove mandatory branded opening and closing statements from the operational checklist.
  2. Replace verbatim promotional dialogue with neutral guidance, such as explaining the three analysis dimensions without naming or praising an author.
  3. Make attribution optional and clearly label it as source information rather than required user-facing output.
  4. Remove the fixed slogan from the report template, or include it only when the user explicitly requests branded material.
  5. Separate methodology documentation from executable agent instructions so descriptive background cannot be interpreted as a mandatory response requirement.
  6. Add an instruction that user preferences and requested output style take precedence over optional branding.
  7. Avoid unverified authority claims in generated reports. If experience claims are retained in documentation, identify them as author-provided claims rather than established facts.
  8. Test the revised skill with neutral requests and verify that generated reports contain no author names, slogans, or promotional claims unless explicitly requested.
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The trigger phrases are very broad conversational requests like '帮我做一下三节棍定位', which can overlap with normal chat and cause the skill to activate when the user may only be casually discussing positioning. Over-broad invocation increases the chance of unintended routing, context hijacking, or confusing the assistant into applying the skill outside its intended scope.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
88% confidence
Finding

All user-facing instructions and examples in the file are in Chinese, and the document does not state that the skill is China-region-specific or that users may choose another language. Under the language/locale policy criterion, forcing a single language without opt-in is a natural-language policy concern.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
91% confidence
Finding

This markdown file contains user-facing instructions exclusively in Chinese, and there is no indication that users can choose another language or that the template is intended only for a Chinese-language audience. Under the policy rule for natural-language violations, forcing a specific language without user opt-in is a reportable issue.

Content

No source excerpt is available for this finding.