Security audit

吃了么

Security checks for vulnerabilities and agentic risk

Overview

This is a coherent food-delivery assistant skill, with expected but privacy-sensitive use of location, preferences, and order history.

Install only if you are comfortable sharing location, cuisine preferences, budget, and order details with the assistant and delivery platforms. Before use, ask the agent not to store history if you do not want local memory, and be cautious about any request to run local scripts because the skill does not define specific allowed commands.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Context-Inappropriate Capability

Medium

Confidence: 92% confidence
Finding: The skill declares use of `exec` for local script execution even though its stated purpose is browsing food-delivery sites, recommending options, and assisting with ordering. Unnecessary local code execution materially expands the attack surface: if later prompts, tool inputs, or local scripts are influenced by untrusted data, this could enable command execution, data access, or lateral actions on the host.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The skill instructs persistent storage of user preferences such as dietary restrictions, cuisine choices, and budget without any user-facing consent or privacy notice. This creates a privacy risk because sensitive behavioral and potentially health-related information may be retained locally without the user's awareness, increasing exposure if the memory store is later accessed or reused unexpectedly.

Missing User Warnings

Medium

Confidence: 89% confidence
Finding: The workflow requests location details and uses them in searches against external food-delivery platforms, but it does not warn users that their location data will be transmitted to third-party services. Location is sensitive personal data; disclosing it externally without clear notice can surprise users and expose them to privacy and tracking risks beyond the local skill environment.

Missing User Warnings

Medium

Confidence: 88% confidence
Finding: The skill assists with ordering and records order history locally, which can reveal recurring habits, location context, spending patterns, and dietary preferences, yet it provides no explicit privacy warning. This is dangerous because purchase history is sensitive behavioral data, and retaining it without clear disclosure or controls increases the risk of unintended profiling or later data exposure.

Static analysis

No suspicious patterns detected.