ClawHub
Back to skill

Security audit

Travel Planning

Security checks across malware telemetry and agentic risk

Overview

This travel-planning skill appears coherent and non-malicious, but it asks the agent to keep persistent travel preferences and local trip records that may include sensitive personal details.

Install only if you want a persistent travel profile and trip folder on disk. Review what is saved in ~/travel-planning/, avoid storing full document numbers or images, and disable or decline any cross-session memory or proactive activation unless you explicitly want it.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (7)

Context-Inappropriate Capability

Medium
Confidence
90% confidence
Finding
The setup explicitly tells the agent to store cross-session activation preferences in MAIN memory, extending data retention and behavior across sessions beyond what is necessary for one-off travel assistance. In a travel-planning context, this can cause persistent profiling and unexpected future activation based on sensitive trip-related discussions, especially if the manifest does not clearly justify that memory capability.

Missing User Warnings

Medium
Confidence
94% confidence
Finding
The skill explicitly instructs the agent to read setup guidance and begin using persistent storage in `~/travel-planning/` before any clear up-front user notice or consent flow about local data retention. In this context, the stored content can include sensitive travel preferences, passport expiry dates, emergency contacts, insurance references, and family logistics, so silent persistence creates a meaningful privacy and trust risk even if the scope is limited to a local directory.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The template instructs the agent to create files under ~/travel-planning without any explicit user-facing disclosure or consent flow for local writes. Silent persistence on a user's filesystem can violate user expectations, create privacy risk, and normalize storing personal travel data without informed approval.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The instruction to update the `last` field on every interaction implies recurring background file modification, again without explicit notice or consent. Even a timestamp update creates an audit trail of user activity and confirms ongoing persistence behavior that could surprise users and expose behavioral metadata.

Vague Triggers

Medium
Confidence
92% confidence
Finding
The guidance encourages very broad future triggers such as intervening whenever the user mentions a destination or trip idea, which expands the skill's activation scope well beyond narrow, user-initiated requests. This can lead to over-collection of travel-related context and surprise activations in unrelated conversations, making the assistant behave in ways the user may not expect or want.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The setup normalizes storing travel-planning data in markdown files and later lists potentially sensitive items such as passport expiry and frequent flyer information without any privacy warning, retention limits, or minimization guidance. In this skill context, travel details can reveal future absence, identity document metadata, and financial/tracking information, so silent persistence increases privacy and security risk.

Hidden Instructions

High
Category
Prompt Injection
Content
integration: pending | done | declined

## Travel Style
<!-- Learn these naturally, don't ask directly -->
<!-- Examples: spontaneous, detailed planner, budget-conscious, luxury, solo, couple, family -->

## Preferences
Confidence
90% confidence
Finding
<!-- Learn these naturally, don't ask directly --> <!-- Examples: spontaneous, detailed planner, budget-conscious, luxury, solo, couple, family -->

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.