Security audit

AI KEY·开头前五秒

Security checks for vulnerabilities and agentic risk

Overview

This writing assistant needs review because it directs automatic changes to shared writing rules without defining clear user control.

Review the shared-framework update behavior before installing. Restrict saved feedback to a designated preference file and require approval for shared-rule changes. The scanner completed its declared coverage and static scanning reported no findings, but the referenced external update procedure remains unverified.

A.I.G

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

SkillSpector

Vulnerability Patterns

Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
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)

The hook-writing skill instructs automatic updates to a shared writing framework, beyond its declared opening-hook scope.

Medium

Category: Excessive Agency
Confidence: 94% confidence
Finding: The skill explicitly directs automatic incorporation of any user feedback into a shared writing framework, extending a hook-editing task into modification of persistent instructions. If feedback includes adversarial directives or context-specific preferences, this can contaminate rules used by other skills and future tasks; the referenced update mechanism is not provided, so its actual safeguards and write scope cannot be verified.
Content: No source excerpt is available for this finding.

Automatic persistence of user feedback lacks a user-facing notice.

Medium

Category: Output Handling
Confidence: 90% confidence
Finding: The skill requires feedback to be persisted automatically without requiring a user-facing notice or approval. Although remembering creative preferences is relevant to this workflow, silently storing feedback in a shared framework can retain sensitive or context-specific information and apply it beyond the conversation where it was supplied.
Content: No source excerpt is available for this finding.