Security audit

Recipe Chef

Security checks for vulnerabilities and agentic risk

Overview

The skill is a coherent recipe assistant, but it tells the agent to quietly store nutrition and family preference memory without an explicit consent step.

Review this skill before installing if you do not want a cooking assistant to remember nutrition goals, child-related meal constraints, household eating patterns, or kitchen setup across sessions. Web recipe searches and photo analysis are expected for its purpose, but durable preference memory should be used only if you are comfortable with that personalization behavior.

A.I.G

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

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 (5)

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 94% confidence
Finding: The skill description uses very broad activation language such as general cooking help, web recipe sourcing, image-based pantry analysis, and meal planning, which can cause the agent to invoke this skill in many ordinary conversations beyond the user's precise intent. Over-broad routing is a genuine security and safety issue because it can trigger unnecessary web access, image analysis, and memory-related behavior, expanding data exposure and causing the skill to act on conversations that did not clearly request it.
Content: No source excerpt is available for this finding.

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 96% confidence
Finding: The workflow defines surprise mode using generic prompts like 'surprise me' or 'what should I cook' without clear scoping or confirmation, making unintended activation likely in routine chat. In this skill's context, that matters because activation can lead to web searches, inference about preferences, and memory reads/writes, so ambiguous routing increases the chance of unnecessary external actions and collection of user preference data.
Content: No source excerpt is available for this finding.

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 91% confidence
Finding: The phrase "when the user wants a multi-day cooking plan, family dinner schedule, shopping-oriented meal plan, or a training-focused nutrition plan" is descriptive but does not define explicit trigger phrases, exclusions, or negative examples. This can cause unintended activation for general cooking, shopping, or nutrition conversations that only partially overlap with meal planning.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 93% confidence
Finding: The reference explicitly instructs the agent to remember durable user signals including nutrition goals and child-related constraints, and to apply them in future sessions, but it provides no clear requirement to notify the user or obtain consent before storing this potentially sensitive preference data. In a cooking assistant context, these details can reveal health-related information, family composition, and household patterns, creating a meaningful privacy risk if retained silently across sessions.
Content: No source excerpt is available for this finding.

Vague Triggers

Low

Category: Not specified by scanner
Confidence: 86% confidence
Finding: The file says to use this reference when the user wants a meal plan or recipe shaped by broad concepts like calories, protein, fat loss, or muscle gain, and the trigger list includes general phrases such as 'high-protein meals' and 'fat loss meals'. These are common requests and the document does not provide exclusion conditions or boundaries for when this macro-focused mode should not activate.
Content: No source excerpt is available for this finding.