Skillv1.0.0

ClawScan security

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

Scanner verdict

SuspiciousApr 15, 2026, 8:04 PM

Verdict: suspicious
Confidence: medium
Model: gpt-5-mini
Summary: The skill's declared purpose (text→video demo) mostly matches its runtime instructions, but there are small incoherences—notably inconsistent config-path declarations and instructions to probe local install paths—that warrant caution before installing or granting credentials.
Guidance: This skill appears to do what it says (cloud-based text→video) and only asks for one credential (NEMO_TOKEN). Before installing: 1) Confirm you want uploads and final videos sent to the external domain mega-api-prod.nemovideo.ai and are comfortable with that provider's privacy and retention policies. 2) Consider creating a limited/throwaway NEMO_TOKEN or use the anonymous flow rather than a shared long-lived token. 3) Note the skill asks the agent to probe local install paths (~/.clawhub/, ~/.cursor/skills/) to set an attribution header — if you prefer not to expose filesystem layout, ask the skill author to stop that behavior. 4) The frontmatter mentions a config path (~/.config/nemovideo/) even though the registry listed none; verify whether the skill will read or write files there. If you need higher assurance, request source code or an explicit privacy/security statement from the publisher before granting credentials.

Review Dimensions

Purpose & Capability: noteThe name/description (text-to-video demo) aligns with the network endpoints, upload, export and SSE flows described in SKILL.md and the single required credential (NEMO_TOKEN). However, the skill's frontmatter requests a config path (~/.config/nemovideo/) and asks the agent to detect install path for attribution headers; these filesystem accesses are not strictly necessary to provide text→video functionality and are inconsistent with the registry metadata that lists no required config paths.
Instruction Scope: noteSKILL.md instructs the agent to make multiple network calls to mega-api-prod.nemovideo.ai (session creation, SSE chat, upload, export polling) and to upload user files to that third-party service — consistent with a cloud render pipeline. It also directs the agent to generate a UUID for anonymous auth if no NEMO_TOKEN is present. The potentially unexpected instruction: detect local install path (~/.clawhub/ or ~/.cursor/skills/) and include that in attribution headers (reads filesystem). This is beyond strictly required video-generation behavior and should be considered privacy-sensitive.
Install Mechanism: okNo install spec and no code files — instruction-only skill. This minimizes on-disk code risk; there is nothing downloaded or installed by the skill itself.
Credentials: noteThe skill requests a single primary secret, NEMO_TOKEN, which is proportional to a cloud API integration. It also documents a config path in the YAML frontmatter (~/.config/nemovideo/) that the registry metadata did not list — an inconsistency. The SKILL.md will create and use an anonymous token when NEMO_TOKEN is absent; that behavior is plausible but means the agent will call the external auth endpoint and share a generated client id with the service.
Persistence & Privilege: okalways:false and disable-model-invocation:false (normal). The skill does not request permanent presence or permissions to modify other skills or system-wide config. It does instruct normal session creation and polling against the external service, but nothing indicates it persists beyond ordinary session tokens.