Back to skill

Security audit

AI代码体检(看不懂代码也能把关)

Security checks for vulnerabilities and agentic risk

Overview

The scanners mark this skill suspicious because its review workflow may run project code, which can undermine a promised read-only review.

Consider whether you are comfortable with a review workflow that may execute code from the project being reviewed. The scan covered both listed Markdown components, but this finding still warrants caution before installing or following the workflow.

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 (1)

The validation workflow runs project code without warning that execution can violate the promised read-only review.

Medium
Category
Output Handling
Confidence
95% confidence
Finding

The skill directs the reviewer to run imports and minimal reproductions whenever possible, but importing project modules can execute arbitrary initialization code with the agent's permissions. In a potentially untrusted codebase, this could modify files, access credentials, or transmit data, contradicting the promised read-only review; the repeated insistence on runtime validation increases exposure despite its legitimate diagnostic purpose.

Content

No source excerpt is available for this finding.