ClawHub
Back to skill

Security audit

Image Handoff

Security checks for vulnerabilities and agentic risk

Overview

The skill performs a coherent image-generation handoff, but it also tells the agent to keep persistent records of raw prompts and feedback without clear user control or retention limits.

Review this skill before installing if your image prompts may contain private, client, personal, or regulated information. It should be limited to minimal request metadata or changed to ask before storing prompts and feedback, with clear retention and deletion behavior.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Findings (3)

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill explicitly instructs persistent logging of prompts, model usage, results, and user feedback to a memory file without any stated minimization, retention limit, access control, or user notice. Image prompts and feedback can contain sensitive personal, commercial, or regulated information, so retaining them by default increases privacy exposure and downstream leakage risk.

Natural-Language Policy Violations

Low
Confidence
81% confidence
Finding
The example handoff payload embeds a specific personal name and locale-script representation, which normalizes forwarding identity attributes that may not be needed for image generation. While presented as an example, it can encourage implementations to include unnecessary personal data in inter-agent messages, creating avoidable privacy exposure.

Ssd 3

Medium
Confidence
97% confidence
Finding
The skill directs operators to persist user image requests and feedback in plain language, which creates a durable record of potentially sensitive content beyond immediate operational need. This expands the attack surface for unauthorized access, later misuse, or accidental disclosure through memory inspection, backups, or model context reuse.

Static analysis

No suspicious patterns detected.