Security audit

臺北市公車定點車機

Security checks for vulnerabilities and agentic risk

Overview

This Taipei bus lookup skill is coherent and purpose-aligned, with minor cautions around broad activation keywords and optional repeating reminders.

Installers should expect public transit API network calls and local cache files. Use this for Taipei bus queries, and if enabling cron reminders, review the schedule and set deleteAfterRun to true for one-time alerts.

SkillSpector

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (6)

Intent-Code Divergence

Medium

Confidence: 95% confidence
Finding: The helper is described as a one-shot arrival reminder, but it returns a recurring cron job with deleteAfterRun set to false by default. In practice this can cause repeated notifications every 2 minutes once the condition is met, creating alert spam and potentially leading users to ignore or disable important notifications.

Vague Triggers

Medium

Confidence: 96% confidence
Finding: The skill instructs the agent to use it whenever a user asks about bus-related information, which is an overly broad activation condition. Broad routing can cause the agent to invoke this skill in contexts where it is unnecessary or less appropriate, increasing the chance of incorrect tool use, privacy over-collection, or unsafe cross-skill behavior if other integrated capabilities are later added.

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The listed trigger phrases include very broad, natural-language expressions such as asking where the bus is or whether it is arriving soon. In an agent environment, overly generic triggers can cause unintended activation during ordinary conversation, leading to accidental data retrieval, user confusion, or invocation of the wrong skill when the user did not explicitly intend to use it.

Vague Triggers

Medium

Confidence: 84% confidence
Finding: The trigger list includes broad natural-language phrases such as 'bus ETA', 'bus route', and mixed-language terms that could activate the skill in conversations where the user did not explicitly intend to invoke external transit lookups. In an agent environment, overly permissive triggers increase the chance of unintended tool use, unnecessary data fetches, and confusing or privacy-impacting behavior.

Missing User Warnings

Medium

Confidence: 89% confidence
Finding: The documentation states that the skill requires network access to external APIs, but it does not clearly warn users that queries, stop names, route lookups, or other request-derived data may be transmitted to third-party services. In an agent setting, this can cause unanticipated disclosure of user inputs or behavioral data to external providers without informed consent.

Vague Triggers

Medium

Confidence: 91% confidence
Finding: The documented trigger phrases are broad enough to match ordinary user conversation about buses, including generic terms like '公交车实时', '公交车路线', and 'Taipei bus'. In an agentic system, this can cause unintended invocation of the skill and unnecessary external data access or tool execution when the user did not explicitly intend to use this capability.

Static analysis

No suspicious patterns detected.