Back to skill

Security audit

物理教案设计

Security checks for vulnerabilities and agentic risk

Overview

ClawScan’s verdict is benign with medium confidence: scanners found no AIG findings, but noted lesson-planning scope concerns and the review was partial because two references were missing.

The verdict is benign, but consider whether you are comfortable with the reported student-profile writeback and experiment-procedure scope. Skillspector scanned all 18 bundled components, yet marked analysis partial because two local references were missing; evidence about those references is incomplete.

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
  • 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 (6)

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

The template supplies concrete experimental procedures and safety protocols that the manifest explicitly delegates to the experiment-coach skill.

Medium
Category
Excessive Agency
Confidence
98% confidence
Finding

The reference supplies actionable electrical-experiment instructions and safety checks despite the manifest delegating operational procedures and safety workflows to the experiment-coach skill. These examples can cause the planner to generate operational guidance without the intended specialist review; low-voltage restrictions and repeated teacher-review warnings reduce, but do not eliminate, this physical-safety risk.

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

L90 introduces student-profile writeback, a capability outside the skill's lesson-planning purpose.

Medium
Category
Excessive Agency
Confidence
90% confidence
Finding

The manifest limits the skill to producing teacher-reviewed physics lesson-plan drafts, but L90 instructs it to send mastery and knowledge-point updates to student learning profiles through a teacher_writeback payload. Checking consent and minimizing the payload reduce privacy risks, but do not justify modifying student records as part of lesson-plan drafting.

Content

No source excerpt is available for this finding.