Security audit

日语会话测试批改

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly coherent, but it can batch post student grades and feedback through course-platform APIs without clearly requiring a final user confirmation or dry-run.

Install only if you are comfortable with a grading assistant that may be configured to use course-platform API tokens. Prefer the default CSV/local-report workflow, keep tokens in local config or environment variables, and require a teacher-reviewed preview before any adapter posts grades or feedback to students.

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

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (4)

Context-Inappropriate Capability

Medium

Category: Not specified by scanner
Confidence: 84% confidence
Finding: The manifest frames the skill as a Japanese conversation audio grading tool with transcription, scoring, feedback, and archiving. In SKILL.md, the skill also defines capabilities to pull submissions from course platforms and push per-student feedback via platform adapters using API credentials, which is a broader integration capability not justified by the narrow grading purpose stated in the manifest.
Content: No source excerpt is available for this finding.

Intent-Code Divergence

Medium

Category: Not specified by scanner
Confidence: 90% confidence
Finding: Lines L283-L287 describe the skill as platform-independent and not bound to any platform, but the same section immediately defines Canvas/Moodle/Feishu adapters, token storage rules, and a download layer that calls platform APIs. This is not merely incomplete documentation; it actively contradicts the claim of being unbound from course platforms.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 89% confidence
Finding: The skill describes automatic pushing/exporting of grades and feedback to external platforms, but it does not require an explicit user confirmation step before modifying student-facing records. In an educational context, unintended posting of incorrect grades or feedback can affect student records, privacy, and workflow integrity, especially because this skill performs AI-assisted pre-grading and batch operations.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 75% confidence
Finding: The file mandates Japanese ASR with --language ja and presents the skill entirely as Japanese-only behavior. Although the skill is clearly for Japanese conversation grading, the file does not explicitly state that this locale restriction is intentional and limited to a Japanese-language assessment context, nor does it present it as a user-selected option.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.