Back to skill

Security audit

AI Forensic Audit(AI 内容溯源审计)

Security checks for vulnerabilities and agentic risk

Overview

This skill provides an evidence-review workflow limited to user-supplied materials, with no executable code or hidden access.

Use it for a clearly defined audit and provide only relevant materials, removing credentials and unnecessary personal information from logs. Generic trigger words may activate it for unrelated requests. Treat its reports as evidence-based assistance, not certified attribution or regulatory approval; standards references alone do not demonstrate technical verification.

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
  • 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 (2)

The trigger list includes generic terms such as “审计” (audit) and “溯源” (trace origins), without explicitly restricting those triggers to AI-related requests.

Low
Category
Trigger Abuse
Confidence
87% confidence
Finding

Although the description presents an AI forensic-audit skill, its explicit trigger list includes standalone terms that also apply to financial audits, product traceability, and other unrelated tasks. Without exclusions or an AI-context requirement for those triggers, the skill could activate outside its intended scope.

Content

No source excerpt is available for this finding.

The activation list includes unqualified terms for tracing origins and auditing.

Low
Category
Trigger Abuse
Confidence
90% confidence
Finding

The “When to use” list includes standalone “溯源” (origin tracing) and “审计” (audit), which also apply to many non-AI tasks. Although the purpose describes AI forensic auditing, these activation entries do not require that context, and the boundary statement does not provide concrete exclusions.

Content

No source excerpt is available for this finding.