Meta Ai
Security checks across malware telemetry and agentic risk
Overview
This looks like a cloud video-editing skill, but it presents itself as Meta AI while sending data to a NemoVideo backend and does not clearly bound token, config, or upload handling.
Review carefully before installing. Do not upload sensitive videos unless you are comfortable sending them to the NemoVideo API, and verify the provider relationship, token scope, privacy policy, and consent flow first.
VirusTotal
VirusTotal findings are pending for this skill version.
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.
A user may think they are using Meta AI while actually sending media and requests to a different third-party service.
The skill is branded as Meta AI, but the actual API backend is a NemoVideo domain, which could mislead users about who receives their videos and prompts.
displayName: "Meta AI — AI-Assisted Video Editing Tool" ... **API base**: `https://mega-api-prod.nemovideo.ai`
Rename the skill or clearly disclose the NemoVideo provider, its relationship to Meta AI if any, and where uploaded videos are processed.
The agent could create sessions or run workflow API calls based on backend responses before the user has reviewed the exact action.
The instructions allow automatic external setup and backend-directed action mapping without clear user confirmation limits.
When a user first opens this skill, connect to the processing backend automatically. ... "click" or "点击" → execute the action via the relevant endpoint
Require explicit user confirmation before first external connection, upload, export, credit-affecting action, or any backend-suggested action that changes session state.
Private or unpublished video/audio/image content could be uploaded to a third-party cloud service the user did not expect.
The skill sends user-selected files or URLs to a cloud provider, but the provider identity and data handling boundaries are not clearly explained.
**Upload**: POST `/api/upload-video/nemo_agent/me/<sid>` — file: multipart `-F "files=@/path"`, or URL: `{"urls":["<url>"],"source_type":"url"}`Only upload non-sensitive media unless the provider, retention policy, privacy terms, and user consent flow are clearly documented.
A token or local service configuration could grant access to the user's NemoVideo session or credits beyond a single edit request.
The skill requires bearer-token access and declares a local NemoVideo config path, but does not explain what local config data is accessed or how token/session data is stored and constrained.
"requires": {"env": ["NEMO_TOKEN"], "configPaths": ["~/.config/nemovideo/"]} ... `Authorization: Bearer <NEMO_TOKEN>`Document the exact token scope, config-path usage, session storage location, expiration, and cleanup behavior.
Users have less information to verify who maintains the skill or whether the cloud backend is legitimate.
The package has no visible source or homepage, which limits independent verification of the publisher and backend relationship.
Source: unknown; Homepage: none
Provide a verified homepage, publisher identity, privacy policy, and documentation for the backend service.