Pilot Service Agents Reference

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:43 PM UTC
• Version:
  1.0.0

Overview

This is a disclosed read-only lookup skill, though its catalog is broader than the short description and queries go through external Pilot service agents.

Install only if you trust the Pilot Protocol setup and are comfortable sending lookup queries through its network and listed service agents. Check each agent's /help output before use, especially for Wikidata, country lookup, weather, random-user, live deals, and Gemini summary flows, and do not submit secrets or sensitive personal/business data.

SkillSpector

By NVIDIA

Vulnerability Patterns

• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (2)

Description-Behavior Mismatch

Medium

• Confidence: 90% confidence
• Finding: The skill’s stated scope is low-stakes reference lookups, but the catalog includes materially different capabilities such as live deals, weather, time, user profile generation, country/entity lookup, and product/book/doc search. This mismatch can cause downstream agents or users to invoke the skill in contexts they were explicitly told to avoid, weakening routing boundaries and increasing the chance of inappropriate data access or misleading outputs.

Intent-Code Divergence

High

• Confidence: 97% confidence
• Finding: The documentation explicitly says not to use this skill for detailed country profiles and knowledge-graph entity lookups, yet the same file advertises `restcountries-name` and `wikidata-wbgetentities`. This contradiction undermines trust in the routing policy and can lead callers to use disallowed capabilities through a supposedly out-of-scope skill, defeating guardrails and policy separation.

VirusTotal

65/65 vendors flagged this skill as clean.