ClawHub
Back to skill

Security audit

content-template

Security checks across malware telemetry and agentic risk

Overview

The skill is mostly a coherent content-template tool, but it needs Review because it can render stored Jinja2 templates with unsandboxed Python power and send rich business context to external LLM routing without clear user-facing controls.

Install only if you trust the templates and the local shared OpenClaw modules it imports. Treat Jinja2 template content as executable-risk input, and do not pass confidential brand profiles, persona data, or knowledge-base excerpts to the generate action unless your tenant policy allows those details to be sent through the configured LLM providers.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (2)

Description-Behavior Mismatch

High
Confidence
94% confidence
Finding
The skill advertises template management and polishing, but also sends rich user and tenant context to external LLM providers to generate full content. This expands the trust boundary significantly and can leak proprietary brand profiles, persona data, knowledge-base excerpts, and operational context to third-party services without the narrower scope implied by the skill description.

Context-Inappropriate Capability

Medium
Confidence
82% confidence
Finding
The template skill invokes a separate script for polish operations, creating an additional execution boundary and increasing attack surface. Even without shell injection, this design makes security depend on another executable component's argument handling, output parsing, timeout behavior, and any network or model calls it performs.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.