Skillv1.0.0

ClawScan security

Ai Video Generator Free Offline · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

SuspiciousApr 21, 2026, 7:07 PM

Verdict: suspicious
Confidence: high
Model: gpt-5-mini
Summary: The skill advertises “offline” use but its instructions require uploading your files and tokens to a remote cloud API (mega-api-prod.nemovideo.ai); that mismatch and the unknown source raise privacy and trust concerns.
Guidance: This skill is misleadingly labeled 'offline' but actually sends your files and session tokens to a remote API (mega-api-prod.nemovideo.ai). Before installing or using it: (1) do not set a permanent NEMO_TOKEN in global environment variables unless you trust the service — prefer ephemeral anonymous tokens if possible; (2) do not upload private/confidential media to this skill unless you verify the service owner's identity, privacy policy, and security practices (there's no homepage or known source listed); (3) if you must try it, run it in a controlled environment (isolated account, small test files) and monitor outbound network traffic; (4) if you require truly offline processing, do not use this skill. The primary red flag is the contradiction between the advertised offline promise and explicit cloud upload/render instructions.

Review Dimensions

Purpose & Capability: concernThe name/description emphasize offline/no-internet usage, but the SKILL.md repeatedly instructs the agent to call cloud endpoints (e.g., mega-api-prod.nemovideo.ai), upload media, create sessions, and start cloud GPU render jobs. This is a clear mismatch between claimed purpose (offline) and actual capability (cloud service). Metadata also references a local config path (~/.config/nemovideo/) and detecting install paths, which is unnecessary for the advertised offline claim.
Instruction Scope: concernRuntime instructions direct the agent to: POST to anonymous-token and session endpoints, upload potentially sensitive files (up to 500MB), include Authorization and attribution headers on every request, and persist session IDs/tokens. Those actions are coherent for a cloud video service but conflict with the UI/language claiming 'offline' processing. Uploading user files to a remote service is explicit here and may expose private content. The instructions also tell the agent to read this file's YAML frontmatter and detect install paths for attribution — a minor scope creep but not necessary for core functionality.
Install Mechanism: okThis is an instruction-only skill with no install spec and no code files, so nothing is written to disk by an installer. That minimizes install-time risk. There are no external download URLs or package installs.
Credentials: noteThe skill requests a single credential, NEMO_TOKEN, which is proportionate to a cloud API client. However, the service endpoint and ownership are unknown (no homepage, unknown registry owner), and the skill offers an anonymous-token flow that issues temporary tokens — meaning the agent will call an external endpoint to obtain credentials. Because the endpoint and project are not verifiable here, handing a token or uploading sensitive media to that service carries privacy/trust risk.
Persistence & Privilege: okThe skill does not request always:true and uses normal autonomous invocation. It instructs saving session_id and tokens for job management, which is expected for a session-based cloud render workflow. It does not request elevated system-wide privileges or to modify other skills.