ClawHub
Back to skill

Security audit

Automation Recipe Book

Security checks across malware telemetry and agentic risk

Overview

This is a coherent automation skill, but it gives agents broad execution and recipe-management authority with under-scoped safeguards for installs, imports, saved files, scheduled actions, and public posting.

Review this before installing if you plan to use it on real workflows. Only run recipe import, community install, rollback, enable, or publishing actions after checking the target files and source. Keep webhook URLs and tokens out of recipes and logs, and prefer dry-run or preview steps before enabling scheduled automations.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (4)

Vague Triggers

Medium
Confidence
90% confidence
Finding
The activation guidance is extremely broad ('use when improving efficiency, automation, batch processing, workflow optimization'), which can cause the skill to trigger in many ordinary requests unrelated to explicit automation consent. In a skill with exec capability and documented file-writing/import/install actions, over-broad invocation increases the chance of unintended operational behavior or state changes.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill documents export, import, and community install operations without warning that they can write files, alter installed recipes, or introduce untrusted content into the workspace. In context, these are direct state-changing actions and, especially for community installation and zip import, can expose users to unsafe modifications or supply-chain-style risks if performed without explicit consent and validation.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The example generator flow writes a YAML file into the user's workspace automatically, but the documentation does not warn that a file will be created before the save occurs. In an exec-capable skill, silent or assumed file creation can lead to unintended workspace modification, overwriting artifacts, or normalization of write actions without informed user approval.

Missing User Warnings

Low
Confidence
79% confidence
Finding
Referencing webhook configuration via environment variables without any privacy or secret-handling guidance is a genuine security hygiene issue. While not an exploit by itself, it increases the likelihood that users expose credentials in plaintext docs, logs, shell history, or shared recipe files when setting up integrations.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.