ClawHub
Back to skill

Security audit

Pilot Service Agents Reference

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed read-only gateway to small online lookup agents, with some scope ambiguity but no hidden code or destructive behavior.

Install this only if you intend to use Pilot Protocol/pilotctl to query remote lookup agents. Check each agent's /help output before sending data, avoid private or sensitive queries, and use the finance, data, or knowledge skills when you need those more specialized domains.

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

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The documentation explicitly tells consumers not to use this skill for detailed country profiles or knowledge-graph entity lookups, yet the listed agents include restcountries-name and wikidata-wbgetentities, directly contradicting those restrictions. Such contradictions create policy bypass opportunities where an agent can access forbidden data domains through a skill that claims they are unavailable, undermining routing and safety controls.

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 presented as a narrow, low-stakes reference lookup, but the embedded catalog expands into materially different capabilities such as live deals, weather, random user profile generation, and broad search/knowledge endpoints. This mismatch can mislead an orchestrator or user into invoking higher-risk or out-of-scope services under a trusted, low-risk label, weakening tool-selection guardrails and data-handling assumptions.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.