Back to skill

Security audit

veterinary-appointment-reminders

Security checks for vulnerabilities and agentic risk

Overview

The bundled skill has limited, ordinary local access, but its script checks fields instead of producing the promised appointment reminder.

Expect to draft the actual reminder yourself and reconcile the template fields with the script. The script does not enforce medical-content restrictions. The optional Windows app is unnecessary for this workflow and needs separate review; the linked [publisher repository](https://github.com/rapidworker1/Multiple-AI-models-Toolkit) describes a broader AI application.

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
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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 (3)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The declared core capability is reminder drafting, but the supplied script only checks supplied field lengths and completeness. Its output contains metadata and validation results, not an owner-facing reminder. Reading an input file and writing a local report are ordinary supporting implementation details; no unrelated resource access or additional capability is evident.

Content

No source excerpt is available for this finding.

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: file_write.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

The skill produces a field-validation report rather than drafting an appointment reminder.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The manifest promises to draft veterinary appointment reminders with preparation instructions, location, and policy notes. At L093–L105, the code emits only field names, character counts, and validation statuses; L107–L117 adds completeness information and generic guardrails before saving the report as vet-reminder.md. The supplied appointment details are never included in an owner-facing reminder.

Content

No source excerpt is available for this finding.