Security audit

Wip Xai Grok Private

Security checks across malware telemetry and agentic risk

Overview

This is a coherent xAI/Grok API wrapper that needs an API key and sends user-selected queries, prompts, and media inputs to xAI, with privacy cautions but no evidence of hidden persistence, destruction, or credential theft.

Install only if you are comfortable using an xAI API key and sending selected searches, prompts, image inputs, and video requests to xAI. Prefer a scoped or revocable key with spending limits, avoid secrets or private media in prompts and file paths, and use XAI_API_KEY directly if you do not want the 1Password lookup.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (5)

Context-Inappropriate Capability

Medium

Confidence: 98% confidence
Finding: When the image argument is not a URL or data URI, the function reads an arbitrary local file path from disk and prepares it for upload. In an agent context, this expands the skill from media editing into local file access, enabling unintended disclosure of sensitive files if a caller supplies a path such as a secret, config, or private document.

Missing User Warnings

Medium

Confidence: 86% confidence
Finding: The README states that the tool will automatically resolve an API key from 1Password or environment variables, but it does not prominently warn users that running the tool may trigger access to sensitive local credentials. In an agent/skill context, implicit credential discovery is more dangerous because users may authorize repo usage without realizing the skill can probe secret sources and use those credentials for outbound API calls.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The skill explicitly supports web and X search via the external xAI API, but it does not warn that user queries and related search parameters are transmitted off-system to a third-party service. This can cause unintentional disclosure of sensitive prompts, internal project names, credentials pasted into queries, or other confidential context when users assume searches are local or privacy-preserving.

Missing User Warnings

Medium

Confidence: 97% confidence
Finding: The image/video generation and editing sections describe sending prompts, image URLs, file paths, base64 media, and seed images to xAI, but they omit any warning that this content leaves the local system and is processed by an external provider. This is risky because users may provide sensitive files or internal media assets without realizing they are being transmitted to a third party.

Missing User Warnings

Medium

Confidence: 97% confidence
Finding: Local file contents may be silently read and then sent to the remote xAI API during image editing, without any explicit notice at the point of use. In agent workflows, this creates a meaningful data-leak risk because a caller may believe they are supplying an image identifier while the skill interprets it as permission to access local disk.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

Detected: suspicious.dangerous_exec, suspicious.env_credential_access

Shell command execution detected (child_process).

Critical

Code: suspicious.dangerous_exec
Location: core.mjs:28

Environment variable access combined with network send.

Critical

Code: suspicious.env_credential_access
Location: core.mjs:22