ClawHub
Back to skill

Security audit

Pilot Service Agents Government

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed directory-style wrapper for querying public government and civic data agents, with some catalog scope inconsistencies but no hidden, destructive, or credential-seeking behavior.

Before installing, confirm you are comfortable using Pilot Protocol and joining its data-exchange network. Treat the listed agent catalog as broader public-data access rather than strictly government-only, and use /help on each agent to verify its exact fields and source before relying on results.

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)

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill metadata and narrative frame this as a government/civic-data category, but the documented catalog materially expands into unrelated or loosely related domains like weather, Nobel prizes, and World Bank datasets. This scope drift can mislead downstream agents or users into invoking broader data sources than intended, weakening policy routing, trust boundaries, and least-privilege assumptions around skill selection.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The documentation states that coverage is country-specific, yet the listed agents include global and cross-country datasets such as Nobel Prize and multiple World Bank sources. This contradiction increases the chance of incorrect agent selection, policy bypass by category confusion, and user reliance on inaccurate operational guarantees.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.