ClawHub
Back to skill

Security audit

Strategic Execution 3D9

Security checks for vulnerabilities and agentic risk

Overview

The business-coaching skill asks agents to save real organizational failure patterns in its reusable instructions without requiring approval or removing confidential details.

Before using confidential business cases, remove the instruction to save actual failure patterns or require explicit approval for anonymized lessons stored in a private location. The remaining workflow is a conventional coaching framework.

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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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 (1)

Instructs agents to persist real-use failure patterns without warning users.

Medium
Category
Output Handling
Confidence
90% confidence
Finding

The instruction to add entries 'after each real use' and 'paste the actual failure pattern' encourages retaining user-derived information in a reusable skill document without consent or sanitization requirements. Because this skill diagnoses internal organizational failures, those entries could expose confidential business or personnel information in later sessions or if the document is shared. This is a conditional persistence risk, not evidence of automatic external transmission or malicious intent.

Content

No source excerpt is available for this finding.