ClawHub
Back to skill

Security audit

blooming-elf 绽放精灵

Security checks for vulnerabilities and agentic risk

Overview

This is a coherent plant-care assistant that stores plant records, reminders, and logs in expected places, with no evidence of deception, exfiltration, or destructive behavior.

Install if you are comfortable with a plant-care skill keeping persistent plant records, reminder settings, local paths or note IDs, and using weather/location context. Ask it where it will store files before onboarding, avoid putting unrelated private details in the plant notes, and only request document links when you intend to share or open that record.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Context-Inappropriate Capability

Medium
Confidence
90% confidence
Finding
The skill includes a workflow to retrieve and provide a shareable document URL for user plant records. Generating or exposing a share link expands access beyond the local assistant context and can leak personal notes, routines, or paths if the link is copied, logged, or sent to the wrong party.

Vague Triggers

Medium
Confidence
85% confidence
Finding
The trigger list contains broad, everyday phrases such as generic help or status expressions that may activate the skill outside clear user intent. In this skill, unintended activation matters because the workflow can read persistent memory, inspect local/IMA documents, and potentially write logs or create automation tasks after partial conversational matches.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill directs automatic creation and modification of local files as part of onboarding and routine use, but it does not require a clear, up-front warning and consent before writing to the filesystem. Unexpected file writes can expose privacy-sensitive data, clutter user systems, or overwrite expected locations if path handling is weak or assumptions about the working directory are wrong.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The skill stores persistent configuration in MEMORY.md, including city, climate, home environment, reminder time, plant list, and document identifiers/paths, without mandating an explicit privacy disclosure and consent step. This creates avoidable privacy risk because the collected data can reveal lifestyle patterns and storage locations, and later workflows depend on repeatedly reading that file.

Static analysis

No suspicious patterns detected.