Security audit

绿灵 · Blooming Elf

Security checks for vulnerabilities and agentic risk

Overview

This is a plant-care assistant that uses local/IMA notes and reminders in a disclosed, purpose-aligned way, though its activation phrases are broad.

Install if you are comfortable with the skill saving plant-care details locally or in IMA notes and using weather/plant lookups. Review ~/.workbuddy/MEMORY.md and any created plant files or reminders if you want to audit or remove saved data.

SkillSpector

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The trigger list includes very broad everyday phrases such as '帮我看看', '状态不好', '太热了', and especially the fallback for unconfigured users that activates on '植物相关话题'. This can cause unintended activation outside clear user consent boundaries, leading the skill to read or update MEMORY.md, local files, notes, or automations in situations where the user may have meant ordinary conversation rather than skill execution.

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The rule for unconfigured users says the skill should trigger on first dialogue or any plant-related topic when no profile exists, which makes the activation boundary unclear and overly expansive. In this skill's context that is more dangerous because activation can immediately lead to reading persistent memory, creating local files or notes, and prompting for automation setup, all of which are side effects that should require clearer invocation.

Static analysis

No suspicious patterns detected.