Back to skill

Security audit

Commute Traffic

Security checks for vulnerabilities and agentic risk

Overview

This skill coherently checks route traffic through TomTom, with expected use of a Python script, a TomTom API key, and user-provided route locations.

Install only if you are comfortable providing a TomTom API key and sending route origins and destinations to TomTom for traffic lookup. Treat it as Switzerland-biased because geocoding is hardcoded to countrySet=CH, even though the prose says any route.

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
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Rogue AgentSelf-Modification, Session Persistence
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (7)

Session Persistence

Medium
Category
Rogue Agent
Content
1. Go to [developer.tomtom.com](https://developer.tomtom.com)
2. Register for a free account
3. In the dashboard, go to **My Apps** → create an app (or use the default one)
4. Copy your API key

**Free tier:** 2,500 requests/day (~75,000/month). Each traffic check uses 3 requests (geocode origin + geocode destination + route), so you get ~830 checks/day for free.
Confidence
60% confidence
Finding
Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Lp3

Medium
Category
MCP Least Privilege
Confidence
93% confidence
Finding
The skill declares executable and data-access capabilities via metadata (python3, environment variable use, and implied network access to TomTom) but does not explicitly constrain tool scope with permissions or allowed-tools. This creates an authorization ambiguity where the runtime may permit broader execution or network behavior than users and reviewers expect, increasing the risk of overreach if the skill is modified or invoked in an unsafe context.

Vague Triggers

Medium
Confidence
90% confidence
Finding
The description includes broad, everyday phrases such as asking about traffic, commute time, best time to leave, or driving conditions, which can match many normal conversations. Over-broad triggering can cause unintended invocation, leading to unnecessary collection and transmission of origin/destination data or unexpected external API calls without clear user intent.

Missing User Warnings

Medium
Confidence
97% confidence
Finding
The skill instructs the agent to send user-provided origin and destination data to the TomTom API but does not warn the user that location data will leave the local environment. Because route endpoints can reveal sensitive places such as home, work, or medical locations, the lack of disclosure undermines informed consent and creates a privacy risk even when the API use is functionally necessary.

External Transmission

Medium
Category
Data Exfiltration
Content
## Geocoding API

**Endpoint:** `GET https://api.tomtom.com/search/2/geocode/{query}.json?key={key}&countrySet=CH&limit=1`

**Example response (abridged):**
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

External Transmission

Medium
Category
Data Exfiltration
Content
## Geocoding API

**Endpoint:** `GET https://api.tomtom.com/search/2/geocode/{query}.json?key={key}&countrySet=CH&limit=1`

**Example response (abridged):**
Confidence
50% confidence
Finding
Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Natural-Language Policy Violations

Medium
Confidence
95% confidence
Finding
The geocoding request forces `countrySet=CH`, which restricts results to Switzerland regardless of the user's requested locations. This is a locale constraint embedded in behavior, and the file does not present it as a user choice or clearly justify the tool as Switzerland-only.

Static analysis

No suspicious patterns detected.