Back to skill

Security audit

物理解题教练

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this physics coaching skill benign with medium confidence: it found no AIG findings, while Skillspector noted some scope concerns and one unresolved reference.

The verdict is benign, not a guarantee of zero risk. Consider whether the broader revision and experimental-method coaching fits your needs. Skillspector inspected all 18 listed components, but overall analysis was marked partial because a local reference in SKILL.md could not be matched to a bundled artifact.

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
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (8)

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 exam-cramming template activates without a concrete problem and provides broad revision instead of coaching the current problem.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The manifest restricts activation to requests containing a specific problem or physical quantity and limits coaching to the current problem. L322–L327 instead accept general exam-preparation requests, and L344–L348 prescribe topic review, targeted practice, or complete systematic training; L405–L409 further introduce three new practice problems while explicitly bypassing the complete four-step workflow.

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.

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 questioning templates teach experimental methods and error analysis that the manifest explicitly excludes.

Medium
Category
Excessive Agency
Confidence
94% confidence
Finding

L353-L356 direct the coach to discuss ammeter internal versus external connection, their error sources, and averaging repeated measurements. These are experimental-method, error-analysis, and data-processing topics that the manifest delegates to the physics experiment coach; they also conflict with this guide’s explicit routing boundary at L587-L595.

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.