ClawHub
Back to skill

Security audit

red-eye-flight

Security checks for vulnerabilities and agentic risk

Overview

This flight-search skill is review-worthy because it can make the agent install and run a third-party travel CLI and send trip details externally without clear consent boundaries.

Install only if you are comfortable with a global npm package being added and with your travel search details being sent to FlyAI/Fliggy. Prefer approving each install and command manually, verify the package publisher and version first, and avoid letting this skill handle unrelated general flight requests.

SkillSpector

Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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)

Vague Triggers

Medium
Confidence
71% confidence
Finding
The activation trigger includes the broad phrase `book a flight`, which can cause the skill to activate for many ordinary travel requests outside its stated red-eye scope. In a skill that mandates external command execution and package installation, overbroad activation increases the chance of unnecessary CLI use, data disclosure to a third party, and unneeded system modification.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill instructs the agent to install and run an external CLI (`npm i -g @fly-ai/flyai-cli`) without any user-facing consent or warning. This is dangerous because it normalizes system modification and execution of third-party code in response to a content file, which could expose the host to supply-chain risk or violate execution boundaries.

Missing User Warnings

Medium
Confidence
92% confidence
Finding
The skill requires obtaining results from the `flyai` CLI, which implies sending travel queries to an external service, but it does not disclose this privacy implication to the user. Users may unknowingly share itinerary details, dates, and locations with a third party, creating unnecessary privacy and compliance risk.

Ssd 4

Medium
Confidence
96% confidence
Finding
The instruction chain strongly pressures the agent to install and execute an external CLI before answering and explicitly forbids safer alternatives. That creates a behavioral override pattern where the model is steered into running unverified commands from skill content, increasing the chance of command execution abuse, supply-chain compromise, and bypass of normal safety expectations.

Static analysis

No suspicious patterns detected.