Back to skill

Security audit

photo-archive-curator

Security checks for vulnerabilities and agentic risk

Overview

This offline photo organizer needs review because its dry-run and paired-file safeguards do not match its code.

Use only on a backed-up copy of your archive until these safeguards are fixed. Even --scan overwrites a report, and applying organization can mishandle paired videos and RAW files. Do not rely on the advertised quarantine or deletion workflow.

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)

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.

Files documented as sidecars are independently renamed and moved even without a matching still.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The organizer includes MOV, MP4, and DNG files as independent candidates and only skips independent processing when a same-stem lowercase .jpg exists, despite documenting these formats as movable only alongside a renamed still. With --apply, unmatched files can therefore be renamed and relocated unexpectedly, potentially disrupting photo/video associations or external references across a large archive; dry-run mode and the explicit apply requirement reduce the immediate risk.

Content

No source excerpt is available for this finding.

Scanning writes and overwrites a report without --apply despite the documented dry-run default.

Medium
Category
Excessive Agency
Confidence
97% confidence
Finding

Running --scan unconditionally truncates and rewrites _scan_report.json without --apply, contradicting the advertised dry-run default and potentially destroying existing content. If that path is an attacker-controlled symbolic link, write_text follows it and can overwrite another file accessible to the process; local, offline operation does not prevent this filesystem risk.

Content

No source excerpt is available for this finding.