Security audit

AI营养师

Security checks for vulnerabilities and agentic risk

Overview

This nutrition skill appears purpose-built and not malicious, but it stores health-related profile data and requests broad agent tools without enough user controls.

Install only if you are comfortable storing nutrition and health-adjacent profile details locally. Use it on a trusted device, review or delete user_data/profile.json when no longer needed, and prefer a future version with explicit save consent, a delete/reset flow, narrower tool permissions, and clear limits on web research sources.

SkillSpector

Vulnerability Patterns

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
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Context-Inappropriate Capability

Medium

Confidence: 93% confidence
Finding: Granting Bash execution to a nutrition-advice skill is broader than necessary for the documented workflow and creates avoidable local-execution risk. If abused, shell access could read or alter local files, chain with stored user profile data, or invoke network/system utilities outside the intended nutrition-reporting scope.

Context-Inappropriate Capability

Medium

Confidence: 93% confidence
Finding: Granting Bash execution to a nutrition-advice skill is broader than necessary for the documented workflow and creates avoidable local-execution risk. If abused, shell access could read or alter local files, chain with stored user profile data, or invoke network/system utilities outside the intended nutrition-reporting scope.

Vague Triggers

Medium

Confidence: 86% confidence
Finding: The trigger list includes very broad everyday phrases such as '怎么吃' and '吃什么', which can cause accidental invocation in unrelated conversations. Over-broad activation is dangerous because it increases the chance the skill collects health-related profile data, writes files, or uses elevated tools when the user did not intend to engage this skill.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill stores sensitive user profile information, including age, sex, height, weight, diet, allergies, and activity level, in a local JSON file without a clear privacy notice, retention policy, or consent flow. Because this is health-adjacent personal data, undisclosed storage materially increases confidentiality and compliance risk if other tools, skills, or local users can access the file.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The code persists a detailed user nutrition profile to disk in profile.json without any consent flow, retention policy, or visibility to the user. Because this profile can contain sensitive health-related data such as age, weight, allergies, diet preferences, and health conditions, unauthorized local access or accidental reuse across sessions could expose private personal and medical information.

Static analysis

No suspicious patterns detected.