ClawHub
Back to skill

Security audit

Pilot Reputation

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly aligned with reputation trend analysis, but it documents indefinite local peer-data collection despite saying it is not for real-time monitoring.

Review this before installing if your Pilot Protocol peer list or reputation scores are sensitive. Use manual or bounded snapshots where possible, restrict permissions on ~/.pilot/reputation, and delete or rotate old snapshot files rather than leaving continuous collection running indefinitely.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • 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)

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The snapshot command persistently writes local-agent metadata and full peer data to disk under ~/.pilot/reputation/data without warning about retention, sensitivity, or access controls. Reputation history, hostnames, addresses, and scores can reveal network relationships and trust posture, creating privacy and operational exposure if the files are over-collected, left unprotected, or exfiltrated.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The skill manifest explicitly says not to use this skill for real-time monitoring, yet the documented workflow implements an infinite polling loop every 300 seconds. This mismatch can cause operators or agents to deploy the skill in a continuous-monitoring role they were told to avoid, increasing resource consumption, daemon/API load, and unintended long-running data collection.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The workflow example continuously collects peer data and creates timestamped files indefinitely, but provides no warning about persistent background collection, storage growth, or the sensitivity of accumulated reputation data. Over time this can build a detailed intelligence log of peer behavior and network state, increasing harm from compromise or accidental disclosure.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

Labeling the example as 'Continuous reputation tracking' directly contradicts the manifest guidance that this skill should not be used for real-time monitoring. In a tool-executing agent ecosystem, contradictory instructions are dangerous because agents may follow the executable example over the prose restrictions and begin unattended monitoring behavior.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.