Skillv1.0.0

ClawScan security

Text To Video Kiss · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

BenignApr 16, 2026, 7:39 PM

Verdict: benign
Confidence: medium
Model: gpt-5-mini
Summary: The skill's declared purpose (remote text→video generation) matches the credentials and API calls it instructs the agent to make, but it will upload user files and issue network requests to an unvetted backend — review and trust the remote service before use.
Guidance: This skill is internally consistent for a cloud text→video tool, but it relies entirely on a remote backend (mega-api-prod.nemovideo.ai) that has no visible homepage or reputation in the package metadata. Before installing or using it: (1) Decide whether you trust that backend — the skill will upload files and prompts (potentially sensitive) for remote processing. (2) Prefer setting your own NEMO_TOKEN (if you have an account) rather than letting the skill obtain an anonymous token for you. (3) Avoid uploading private or regulated content (PII, secrets, or minors) unless you have reviewed the service's privacy/TOS. (4) Note the skill reads environment and install/config paths to derive headers — this is used for attribution but could reveal platform info. If you need higher assurance, ask the publisher for a homepage or privacy policy and an explicit security/privacy statement before proceeding.

Purpose & Capability: okThe skill claims to call a remote video-rendering backend and requires a single token (NEMO_TOKEN) plus a config path for nemo-related data; these requirements are proportionate to a cloud text-to-video service.
Instruction Scope: noteThe SKILL.md instructs the agent to read NEMO_TOKEN from the environment (or acquire an anonymous token via POST), create sessions, upload files, run SSE interactions, and poll render endpoints. These actions are coherent with the stated purpose, but they involve network activity to mega-api-prod.nemovideo.ai, uploading user files, and checking install/config paths (to derive X-Skill-Platform). The instructions explicitly ask not to expose tokens, which is good, but the agent will transmit user content and metadata to a third-party API.
Install Mechanism: okNo install steps or third-party downloads are present (instruction-only). Nothing is written to disk by an installer; risk is limited to runtime behavior.
Credentials: okOnly a single token (NEMO_TOKEN) and a nemo config path are required, which matches a remote API client. There are no unrelated credential requests or broad environment access declared.
Persistence & Privilege: okThe skill does not request always:true and does not ask to modify other skills or system-wide config. It will create session tokens and use them for API calls, which is expected for a remote service.