Back to skill

Security audit

assistant

Security checks for vulnerabilities and agentic risk

Overview

This is a markdown-only product-support skill with no executable behavior, though users should control any FAQ learning and handle biometric-product information carefully.

Install only if you want a Chinese-led assistant for PalmAI/KYC Standard presales support. Review generated customer replies for privacy, consent, and authorization language, especially around palm biometric data, logs, exports, remote upgrades, and resets. Do not allow automatic FAQ updates to store customer secrets, personal data, or unsupported commitments.

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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill description defines a broad, unclear activation scope ('as an assistant' and for autonomous knowledge accumulation) without precise boundaries or triggers. This can cause unintended invocation or over-selection of the skill, leading to prompt routing mistakes, scope creep, or the skill being used in contexts it was not designed to handle.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The skill mandates bilingual output for all cases without user choice, which can override caller preferences and create unnecessary disclosure or formatting problems in downstream workflows. In multi-agent or enterprise settings, rigid output constraints can reduce controllability and cause accidental leakage of internal context into customer-facing text.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The entire FAQ and instructional content is presented only in Chinese, with no indication that users may choose another language or that the skill is intentionally limited to a Chinese-speaking audience. 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.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

This markdown file describes collection and registration of palm biometric data, automatic log upload, record upload, remote upgrade, reset, and data export features, but it does not include any user-facing warning about impacts to personal data, privacy, or system integrity. Under the markdown criteria for SQP-2, descriptions of behaviours affecting sensitive user data and device state should disclose those risks.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.