Back to skill

Security audit

tingyun-appeal-strategist

Security checks for vulnerabilities and agentic risk

Overview

This skill provides appeal checklists and drafts for human review, but users should limit the personal information they share.

Share a factual summary and relevant redacted excerpts with the assistant. Keep originals secure and submit unredacted evidence only when required through the platform’s official appeal channel. Verify current deadlines and review every draft before submitting.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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 (2)

The skill requests complete original evidence without warning users about disclosing sensitive information.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The skill encourages users to provide complete factual materials and requires uncropped evidence without distinguishing what belongs in the assistant conversation from what should be submitted privately to the platform. In this context, orders, conversations, contracts, and receipts can expose customer identifiers, contact details, or financial information unnecessarily; there is no evidence of deliberate exfiltration.

Content

No source excerpt is available for this finding.

The evidence-submission checklist requires complete screenshots containing personal information without a privacy warning.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

The checklist explicitly requests complete chat histories, order screenshots, buyer addresses, and contact details without privacy safeguards or a relevance limitation. Although authentic evidence is appropriate for official appeals, applying this checklist to assistant uploads or broadly shared working documents could unnecessarily disclose buyers’ personal information and unrelated conversation content.

Content

No source excerpt is available for this finding.