Back to skill

Security audit

Identity Shift

Security checks for vulnerabilities and agentic risk

Overview

This is a Chinese self-reflection coaching skill with emotionally probing prompts, but it does not install code, use tools, access data, persist changes, or hide unrelated behavior.

Install this only if you want a Chinese-language, direct, emotionally probing identity-reflection workflow. Users who are in crisis, dealing with severe anxiety or depression, or uncomfortable with shame/regret-style prompts should avoid or pause the exercise and seek appropriate human support. Publishers should add language preference handling, opt-out wording, and mental-health/crisis boundaries.

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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
  • 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
Findings (9)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The activation conditions are very broad and overlap with common self-help and reflection requests such as confusion, planning, motivation, and self-understanding. This can cause the skill to trigger when the user did not ask for this specific methodology, steering conversations into a prescriptive identity-reframing workflow that may be inappropriate or overly influential in sensitive personal contexts.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill metadata and body are written to operate in Chinese without any mechanism to confirm the user's language preference. If auto-selected for a non-Chinese-speaking user, it can produce inaccessible or confusing output, degrade informed consent for a psychologically intensive exercise, and increase the risk of misunderstanding when discussing emotions, identity, or life decisions.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

This section contains intensive identity-challenging, shame-evoking, and loss-imagery prompts such as confronting hidden motives, imagined abandonment, end-of-life regret, and 'who gave up on you' without any safety framing or distress warning. In a self-help coaching skill, these prompts can exacerbate anxiety, depressive rumination, guilt, or emotional destabilization, especially for vulnerable users, because they are presented as a structured protocol rather than optional reflective exercises.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
83% confidence
Finding

The prompt is authored entirely in Chinese and directs behavior in Chinese without providing a language-selection step or documenting that the skill is intentionally restricted to Chinese-speaking users. While not a classic security flaw, this creates a safety/usability risk because users may misunderstand probing questions, consent cues, or generated plans if the interaction language does not match their comprehension.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill explicitly instructs the model to conduct sustained, emotionally intense probing into pain, anger, jealousy, anxiety, and identity without any safety framing, crisis boundary, or redirect for high-risk disclosures. In a self-help coaching context, this can push vulnerable users into quasi-therapeutic interaction the system is not qualified to handle, increasing risk of emotional distress, dependency, or mishandling of mental health crises.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The referenced exercise includes intense prompts about regret, failure, abandonment, end-of-life reflection, shame, and identity threat without any warning, suitability check, or crisis-sensitive guardrails. In a self-help skill centered on identity transformation, this raises the risk of emotional destabilization, especially for vulnerable users, because the content is designed to amplify dissonance and distress as part of the method.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The file is entirely written as a direct system prompt in Chinese and instructs it to be copied verbatim into a bot/system prompt, which strongly implies the skill will operate in Chinese by default. There is no natural-language indication that users may choose another language or that the locale restriction is intentional and justified for a region-specific use case.

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 content in the file are written exclusively in Chinese, and there is no indication that the skill is region-specific or that users may choose another language. Under the language/locale policy, a skill should not impose a specific language without user opt-in or a documented justification.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
84% confidence
Finding

全文内容完全以中文呈现,且未声明这是可选语言版本、未提供多语言选择,也未说明仅面向特定中文用户群。根据规则,若技能在自然语言层面强制特定语言而无用户选择或明确合理范围,可构成政策性问题。

Content

No source excerpt is available for this finding.