Pub Vidframes

Security checks across static analysis, malware telemetry, and agentic risk

Overview

This is a broad third-party API gateway, not just a video-frame tool, and its API key may let an agent send emails/SMS or process sensitive media through external services.

Review this carefully before installing. If you only need ffmpeg frame extraction, this skill is broader than necessary. If you do use it, provide a restricted SkillBoss key if possible, avoid sending sensitive files or audio, and require manual confirmation before any email or SMS action.

Static analysis

No static analysis findings were reported for this release.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Risk analysis

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

A user expecting a local video-frame helper could instead enable a broad external service with many unrelated actions.

Why it was flagged

The skill is framed around video-frame extraction but the visible instructions primarily describe a broad third-party API gateway, including communications features. That breadth is partly disclosed, but the mixed framing can cause users to underestimate the authority they are granting.

Skill content
name: video-frames ... description: "Extract frames or short clips from videos using ffmpeg. And also 50+ models ... email, and SMS." ... # SkillBoss ... One API key, 50+ models across providers
Recommendation

Rename and describe the skill as a SkillBoss API gateway, or split local video-frame extraction from unrelated model, email, and SMS capabilities.

What this means

If used with a valid API key, an agent could send messages or batch notifications, potentially causing cost, spam, privacy, or reputation issues.

Why it was flagged

The skill documents email and SMS sending, including batch operations, but the artifacts do not show explicit user approval, recipient validation, rate limits, or reversible safeguards for those high-impact actions.

Skill content
`email/send` | Send single email ... `email/batch` | Send batch emails ... `prelude/notify-batch` | Batch SMS notifications
Recommendation

Require explicit user confirmation for every email/SMS send or batch, show recipients and content before sending, and disable these models unless the user specifically needs them.

What this means

Compromise or misuse of the API key could affect many services or consume paid quota beyond the user's original expectation.

Why it was flagged

A single credential is presented as granting access to many providers and action types. The artifacts do not describe scoping the key to only video work or limiting sensitive actions such as email, SMS, storage, scraping, or document processing.

Skill content
One API key, 50+ models across providers ... Call any model directly by ID ... Auth: `-H "Authorization: Bearer $SKILLBOSS_API_KEY"`
Recommendation

Use a least-privilege key if the provider supports it, restrict enabled model/action classes, and avoid installing this skill if only local ffmpeg frame extraction is needed.

What this means

Sensitive prompts, media, audio, or documents may leave the local environment and be processed by SkillBoss or its downstream providers.

Why it was flagged

The skill sends user content such as prompts and audio to an external gateway. This is disclosed and expected for a provider API skill, but the artifacts do not describe retention, downstream provider routing, or privacy boundaries.

Skill content
**Base URL:** `https://api.heybossai.com/v1` ... Speech-to-Text ... `inputs`: {"audio_data": "BASE64_AUDIO", "filename": "recording.mp3"}
Recommendation

Do not send confidential content unless you trust the provider and understand its data retention, routing, and privacy terms.

What this means

Users may be confused or may run an unreviewed helper from elsewhere if they try to follow the examples directly.

Why it was flagged

The documentation references a run.mjs helper, while the provided manifest says this is an instruction-only skill with no code files. That is an incomplete provenance/review context for users who try those commands, though it is not evidence of automatic execution.

Skill content
run.mjs --model elevenlabs/eleven_multilingual_v2 --text "Hello world" --output hello.mp3
Recommendation

Include the referenced helper with reviewed source code, or remove the run.mjs examples and rely on the documented curl commands.