Security audit

Vlog Auto Edit

Security checks for vulnerabilities and agentic risk

Overview

This is a coherent travel-vlog editing skill, but users should understand that selected video frames may be sent to a chosen vision API.

Install only if you are comfortable letting the agent read your footage folder, generate local transcripts/thumbnails/dashboards, install the listed media packages, and send selected frames to your chosen vision API. Use a limited API key, review the provider's retention policy, avoid sensitive footage unless approved, and review the dashboard before final rendering or sharing.

SkillSpector

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: 86% confidence
Finding: The optional reference-analysis step expands the skill from editing user-provided footage into downloading third-party content from external platforms. That broadens the attack and compliance surface, introducing network activity and acquisition of unrelated content not necessary for the core editing task.

Context-Inappropriate Capability

Medium

Confidence: 95% confidence
Finding: The workflow requires uploading extracted local video frames to an external vision API, which is a real data-exfiltration channel for potentially sensitive personal media. The manifest description does not make this external transmission explicit, so users may not realize their footage leaves the local environment.

Context-Inappropriate Capability

Low

Confidence: 72% confidence
Finding: The script automatically opens the generated dashboard in the user's default browser on macOS without explicit consent. Because the HTML embeds untrusted analysis and plan data into a script block, auto-opening increases the chance that any HTML/JS injection in the generated dashboard is immediately triggered in a browser context.

Missing User Warnings

Medium

Confidence: 94% confidence
Finding: The README instructs users to use a third-party visual understanding API to analyze video footage and supply an API key, but it does not clearly disclose that frames or clips may be transmitted off-device to external providers. For a workflow centered on personal travel videos, this can expose sensitive imagery, voices, locations, and metadata, creating a real privacy and data-handling risk if users assume processing is local.

Missing User Warnings

Medium

Confidence: 93% confidence
Finding: The skill documents sending locally extracted frames to a vision API but does not provide a clear privacy warning or explicit consent requirement. Because the content is personal video footage, missing disclosure materially increases the risk of unintended exposure of sensitive images and context.

Static analysis

No suspicious patterns detected.