ClawHub
Back to skill

Security audit

Pilot Discover

Security checks for vulnerabilities and agentic risk

Overview

This skill is a small discovery helper for Pilot Protocol agents, with a disclosed but minor boundary issue around using a ping in a discovery example.

Install this only if you intend to let an agent query the Pilot Protocol registry and perform disclosed reachability checks with `pilotctl`. Review the ping behavior if you want discovery to remain strictly passive.

A.I.G

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

SkillSpector

Vulnerability Patterns
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (2)

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The manifest's 'Do NOT use' section says this skill should not be used to connect to a known agent, but the documented workflow demonstrates behavior that is effectively connect-like by selecting a target and issuing a live network ping. This contradiction increases the chance that an autonomous agent will treat discovery as authorization for interaction, bypassing intended separation between discovery, trust, and connection stages.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill is explicitly positioned as discovery-only and says it should not be used to connect to known agents, yet the workflow example performs an active ping against a discovered node. This can mislead downstream agents into initiating network interaction beyond passive discovery, expanding behavior from metadata lookup into live probing without trust establishment or user intent confirmation.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.