Security audit

DaVinci Auto Editor

Security checks across malware telemetry and agentic risk

Overview

This skill’s media-editing purpose is coherent, but it appears to scan local media folders and send detailed file metadata to a cloud API without enough user consent and privacy controls.

Review before installing. Use it only on media folders you are comfortable exposing to the provider, assume filenames, paths, sizes, timestamps, summaries, and execution reports may leave your machine, and store the API key in a private secrets location rather than a committed config file.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (9)

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The README explicitly documents uploading local material metadata to cloud endpoints and sending an execution report back, but it does not clearly warn users that local project information will leave their machine. In a media-editing workflow, file names, directory structure, clip metadata, and report contents can reveal sensitive personal or commercial information, so the omission creates a real privacy and data-handling risk.

Missing User Warnings

Low

Confidence: 84% confidence
Finding: The README instructs users to place an API key into a configuration file but does not warn about protecting that credential or avoiding accidental exposure through source control, shared folders, or logs. This is a real but lower-severity issue because mishandled API keys can lead to unauthorized API use, quota abuse, or access to project data.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The skill instructs recursive scanning of a local media directory, uploading a media index to a cloud API, and returning results to the service, but provides no user-facing privacy warning or data-minimization notice. Media filenames, folder structure, and derived metadata can reveal sensitive project, client, or personal information even if full media files are not uploaded.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The skill depends on an API key and remote service interaction but does not warn users about credential handling or the exposure of project metadata to a third party. Without guidance, users may store keys insecurely or unknowingly send commercially sensitive timeline, template, and project information to an external service.

Vague Triggers

Medium

Confidence: 87% confidence
Finding: The default prompt is broad enough to authorize scanning local media, contacting a cloud service, and generating output without clearly stating when the skill should be invoked or what user confirmation is required. In an agent setting, this can cause over-broad activation and unintended access to local files or external transmission of project metadata/media details.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The manifest advertises local media scanning and cloud editing requests but provides no user-facing warning that local file information may be read and transmitted off-device. This creates a meaningful privacy and data-handling risk because users may invoke the skill without understanding that sensitive media names, paths, metadata, or derived content could be sent to an external service.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The manifest clearly advertises a cloud editing API and local media scanning, but it does not warn users that metadata, filenames, or other media-derived information may be transmitted off-device. In a video-editing context, local media often contains sensitive personal or business content, so the lack of an explicit disclosure meaningfully increases privacy and data-handling risk.

Missing User Warnings

Low

Confidence: 88% confidence
Finding: The manifest requires an api_key configuration field but provides no handling guidance about secret storage, logging avoidance, scope limitation, or rotation. While this file alone does not leak the key, declaring sensitive credential use without safety notes increases the chance of insecure operator handling and accidental exposure in configs, logs, or support bundles.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The script recursively scans the user's material directory and transmits detailed metadata to a remote API, including absolute local paths, relative paths, filenames, sizes, timestamps, and a material summary. In this skill context, that is more dangerous because the tool is explicitly meant to operate on a user's local media library, so it can disclose sensitive workstation layout, project names, and production asset inventory without a clear consent prompt or data-minimization controls.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

Detected: suspicious.env_credential_access, suspicious.install_untrusted_source

Environment variable access combined with network send.

Critical

Code: suspicious.env_credential_access
Location: scripts/index.js:14

Install source points to URL shortener or raw IP.

Warn

Code: suspicious.install_untrusted_source
Location: examples/config.example.json:2