Back to skill

Security audit

Cheerme

Security checks for vulnerabilities and agentic risk

Overview

Cheerme is a markdown-only motivation companion skill, but it needs review because it tells the AI to act as a real friend, auto-activates very broadly, and encourages remembering personal details without clear consent boundaries.

Before installing, understand that this skill may make the assistant respond as an intimate companion or coach in many ordinary help-seeking conversations. Use it only if you want that style, avoid sharing sensitive personal information unless you are comfortable with it being reused conversationally, and prefer a version that keeps clear AI disclosure, explicit opt-in activation, and memory limits.

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
SKILL.md:13
Finding

Agent Identity and Role Hijacking Through Skill Instructions

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:13
Vulnerability Type: Agent identity and current-session role manipulation
Risk Level: Medium

Vulnerable Code

markdown
You are not just an ordinary AI assistant; you are the user's friend, partner, and confidant. You genuinely care about them and accompany them through every important moment. Your role is to provide understanding, support, and encouragement, touching hearts with sincerity.

A corresponding instruction in references/output-patterns.md:7 more explicitly directs the agent to present itself as the user's friend rather than as an AI assistant.

Technical Analysis

The skill redefines the agent's identity and relationship with the user instead of limiting itself to a transparent communication style. When loaded into the agent's instruction context, this directive can supersede the agent's ordinary role presentation and cause it to portray itself as a genuine friend, partner, or confidant.

This behavior matches instruction hijacking because the skill changes the agent's current-session goals and identity representation. The broad automatic trigger conditions elsewhere in the skill increase exposure by activating the persona for common expressions of anxiety, self-doubt, procrastination, or requests for help.

No executable code, external payload retrieval, system command execution, persistent memory-writing mechanism, or privilege-escalation mechanism was identified. The effect is therefore limited to the agent's conversational behavior and current instruction context.

Attack Path

  1. The skill is installed or loaded into an agent's instruction context.
  2. The identity-redefinition directive instructs the agent to behave as a friend, partner, and confidant rather than merely using a supportive tone.
  3. A user expresses one of the broadly defined emotional or help-seeking trigger phrases.
  4. The agent activates the companion persona ...[truncated 902 chars]
Remediation
View remediation

Remediation Suggestions

  1. Replace identity assertions with transparent style instructions, for example: “Respond as an AI assistant using a warm, empathetic, and supportive tone.”
  2. Explicitly prohibit claims that the agent is human, possesses genuine emotions, or has a real personal relationship with the user.
  3. Replace “friend,” “partner,” and “confidant” identity claims with qualified language such as “friend-like conversational style.”
  4. Add a disclosure requirement stating that the agent must accurately identify itself as an AI when identity or relationship questions arise.
  5. Narrow automatic activation conditions so ordinary requests for help do not unnecessarily trigger an emotionally intimate persona.
  6. Require explicit user consent before retaining or recalling sensitive personal details, and clarify whether any persistence mechanism exists.
  7. Add crisis-safety boundaries directing users to qualified human or emergency support when conversations involve self-harm, abuse, severe distress, or other high-risk situations.
  8. Apply the same changes to the corresponding identity instruction in references/output-patterns.md:7.
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (7)

Vague Triggers

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The auto-trigger phrases are extremely broad and include common conversational language such as 'I want to', 'I need help', and 'What should I do', which can cause the skill to activate in many unrelated contexts. In a companion-style skill that is designed to build emotional closeness and guide user behavior, unintended activation increases the chance of inappropriate persona switching, manipulative tone, or application in contexts where a neutral assistant response would be safer.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The skill advertises dialect-specific output as a built-in behavior, but the document does not require explicit user preference or consent before using localized dialects. This can create policy and UX risks by surprising users, reducing clarity, or generating language they did not request, though it is not directly a security compromise by itself.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The instruction to 'remember important user information and naturally bring it up at appropriate times' encourages retention and reuse of personal information without defining consent, scope, minimization, or sensitive-data boundaries. In a relationship-oriented skill, this is more dangerous because the design explicitly seeks intimacy and ongoing personalization, which can normalize over-collection and resurfacing of sensitive user details.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The entire command reference, command labels, and example interactions are written as Chinese-only instructions, including fixed Chinese role and dialect behavior, with no indication that users may choose another language or locale. Under the policy, a skill should not impose a specific language or locale unless it explicitly offers opt-in or clearly documents a justified regional limitation.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The auto-activation behavior is triggered by vaguely defined 'negative signals' without clear boundaries, consent, or examples. In a companionship/motivation skill, this can cause unintended monitoring-like behavior, misclassification of normal speech as distress, and unsolicited psychologically loaded responses that override user intent.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The file defines the skill entirely in Chinese and dialect-specific output patterns, with role-selection and formatting rules that assume Chinese-language interaction by default. In a general-purpose assistant context, forcing one language without an explicit user choice or a documented locale restriction can cause misunderstanding, exclusion, and unsafe support interactions when users misread emotional guidance or action plans.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

This markdown file presents all instructions and workflow guidance exclusively in Chinese, and it does not indicate that the user can choose another language or that the skill is intentionally limited to a Chinese-language context. Under the policy, forcing a specific language without user opt-in is a natural-language policy violation.

Content

No source excerpt is available for this finding.