skill-base-cli

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:38 PM UTC
• Version:
  1.0.0

Overview

This skill is a transparent CLI helper for Skill Base; it can install, update, publish, and configure skills, but those actions fit its stated purpose and are disclosed.

Install only if you intend to use the Skill Base CLI. Before running install, update, publish, import, or global npm commands, confirm the target skill, directory, account, and `SKB_BASE_URL`; treat login verification codes and any resulting tokens as sensitive and keep them in the local login flow.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code
• Excessive Agency: Unrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (2)

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The skill explicitly tells assistants to perform actions by running terminal commands, and several documented commands (`install`, `update`, `publish`, and some `list` follow-up flows) can modify local files or configuration. Because the instructions do not require the assistant to warn the user before making filesystem changes, a user may not realize that executing the suggested command will alter installed skills, config state, or local directories.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: The publish and GitHub import sections describe workflows that send skill contents or repository-derived data to a remote Skill Base server, but they do not instruct the assistant to obtain informed user consent or warn about data transmission. This is risky especially because `SKB_BASE_URL` may point to a private or non-default server, so local content could be uploaded to an unintended endpoint.

VirusTotal

67/67 vendors flagged this skill as clean.