Back to skill

Security audit

代码

Security checks for vulnerabilities and agentic risk

Overview

This skill appears intended for IRIS/ObjectScript code formatting, but it asks for broad command and file-editing authority without clear limits.

Review this before installing if you only need a formatter. It may be reasonable for trusted IRIS/ObjectScript repositories, but limit use to code you intend to modify and avoid allowing arbitrary shell commands unless you have a clear, user-approved command list.

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (2)

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill is presented as a code formatter/reviewer, but it declares the powerful exec capability without narrowly constraining when or how commands may be used. In an agent environment, this creates a privilege mismatch: routine user requests about formatting or review could be escalated into shell execution, enabling command injection, unintended file/system changes, or abuse of the host environment.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
80% confidence
Finding

The invocation guidance is broad enough to match common development, debugging, and deployment requests rather than a tightly scoped formatting task. In combination with read/write/exec capabilities, over-broad triggering increases the chance the agent routes unrelated requests into a high-privilege skill, causing unnecessary code modification or command execution in contexts the user did not specifically authorize.

Content

No source excerpt is available for this finding.