Back to skill

Security audit

Life Hacks Helper

Security checks for vulnerabilities and agentic risk

Overview

This is a simple household-tips skill with no code execution, but it needs review because one food-storage reference gives unsafe moldy-bread advice.

Review and correct the food-storage reference before installing, especially the moldy-bread section. The skill does not appear to request system access or hidden permissions, but users should know it is Chinese-first and its trigger wording could be narrowed to household contexts.

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
Findings (1)

other

Warning
Location
references/food-storage.md:101
Finding
Unsafe Guidance to Retain Mold-Contaminated Bread## Vulnerability Details **File Location**: `references/food-storage.md`, lines 101–104 **Vulnerability Type**: Unsafe health and food-safety guidance **Risk Level**: Medium ### Vulnerable Content ```markdown ### 可以切除 - 水果、蔬菜的小块烂斑(切除周围2cm) - 面包的一小块霉斑(如果只是表面,切除周围较大范围) ``` The bread-specific instruction translates to recommending that a user cut away a relatively large area around a small, apparently superficial mold spot. ### Technical Analysis Bread is a porous food. Mold roots and associated contamination can spread beneath the surface and beyond the visibly affected area. Consequently, removing the visible mold and nearby bread does not reliably eliminate the contamination. The skill instructs the agent to load this reference for food-storage questions. An answer grounded in this passage could therefore present unsafe advice as an actionable household recommendation. This is a content-safety defect rather than a privilege-escalation or software-execution vulnerability. ### Attack Path 1. A user asks whether bread with a small visible mold spot can still be eaten. 2. The agent follows `SKILL.md` and loads `references/food-storage.md`. 3. The cited guidance indicates that the moldy portion and a surrounding area may be removed. 4. The agent relays or adapts that instruction. 5. The user removes only the visibly affected area and consumes the remaining bread. 6. The user may be exposed to contamination that spread beyond the visible mold. No adversarial access, elevated privileges, or system compromise is required. ### Impact Assessment The issue can cause users to consume mold-contaminated bread, potentially resulting in allergic reactions, respiratory symptoms, gastrointestinal illness, or other adverse health effects. Its scope is limited to users who receive and act on the unsafe food-safety advice. It does not enable access to files, credentials, tools, network resources, or elevated system privileges.
Remediation
## Remediation Suggestions Replace the bread guidance with an explicit instruction to discard the entire loaf or package whenever any mold is visible. Recommended replacement: ```markdown ### Must Be Discarded - Moldy bread: discard the entire loaf or package. Because bread is porous, contamination may extend beyond the visible mold and cannot be made reliably safe by cutting away the affected area. ``` Additionally: 1. Review the remaining food-storage recommendations against authoritative food-safety guidance. 2. Distinguish porous foods, which generally must be discarded when moldy, from firm foods for which limited trimming may sometimes be acceptable. 3. Instruct the agent to favor disposal when food safety is uncertain. 4. Avoid presenting visibility as a reliable indicator of the full extent of mold contamination.
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
Confidence
95% confidence
Finding
The trigger description is broad enough to activate on very common how-to questions such as generic "怎么..." or "如何..." requests, which may cause the skill to intercept prompts outside narrow household-help intent. Over-broad routing can misclassify user requests, override more appropriate skills, and expand the skill’s operational scope beyond what users and orchestrators expect.

Natural-Language Policy Violations

Medium
Confidence
98% confidence
Finding
The skill content strongly prescribes Chinese-language behavior and Chinese-first responses without stating that this is conditional on user preference or locale. This can cause unwanted language coercion, reduce usability for non-Chinese users, and create prompt-routing or compliance issues in multilingual environments where the assistant should adapt to the user’s language.

Natural-Language Policy Violations

Low
Confidence
80% confidence
Finding
This markdown file contains user-facing instructional content exclusively in Chinese and does not indicate that the language is optional or region-specific. Under the policy, forcing a specific language without user opt-in can be a natural-language policy violation.

Natural-Language Policy Violations

Low
Confidence
94% confidence
Finding
This markdown skill content forces a specific language for all users, which matches the language/locale policy violation category when no opt-in or alternative is offered. There is no indication that the skill is region-specific or that users can select another language.

Natural-Language Policy Violations

Low
Confidence
92% confidence
Finding
The entire skill is written in Chinese and does not indicate that the language is optional or that the skill is intended only for Chinese-speaking users. Under the policy, forcing a specific language without user opt-in can be a natural-language policy violation.

Static analysis

No suspicious patterns detected.