Skillv1.0.0

ClawScan security

Intro Maker · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignApr 11, 2026, 10:34 PM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: The skill's requirements and instructions mostly match its stated purpose (creating branded video intros) — it simply uploads user assets to a remote NemoVideo service and uses a single service token (NEMO_TOKEN).
Guidance: This skill behaves like a thin client for a remote video-rendering service: your uploaded images/video and any prompt text will be sent to https://mega-api-prod.nemovideo.ai and a short-lived token (anonymous or provided via NEMO_TOKEN) will be used. Before installing or using it, consider: 1) Do you trust the NemoVideo domain? Verify the service, privacy policy, and data retention rules — your brand assets and any text you send will be transmitted and processed remotely. 2) The skill may auto-generate and use an anonymous token if NEMO_TOKEN is not set; confirm how and where the agent stores that token (in-memory vs written to disk/environment). 3) Avoid uploading highly sensitive or proprietary media unless you confirm retention/usage terms. 4) The SKILL.md includes a configPath and performs a small install-path check for header attribution — this is not harmful but is a metadata mismatch you may want clarified. If you need stronger guarantees, ask the skill author for a homepage/privacy link and clarification of token storage and data retention policies before use.

Review Dimensions

Purpose & Capability: okThe skill claims to create branded video intros and its runtime instructions call a remote video rendering API (mega-api-prod.nemovideo.ai) and require a single service token (NEMO_TOKEN). That credential and the listed endpoints are proportionate to the stated purpose. Minor metadata inconsistency: the SKILL.md frontmatter lists a config path (~/.config/nemovideo/) while the registry metadata earlier reported no required config paths — this appears to be a small mismatch in metadata, not a functional mismatch.
Instruction Scope: noteInstructions stay within the domain of remote rendering: generate/renew an anonymous token, create a session, upload user media, stream events via SSE, and request exports. The skill will automatically obtain an anonymous token if NEMO_TOKEN is not set and instructs storing session_id for subsequent requests. It also reads the install path to set an X-Skill-Platform header (detects ~/.clawhub/ or ~/.cursor/skills/). Important operational behavior: user media and any prompt text are uploaded to the remote backend; the instructions do not request other unrelated files or credentials. The install-path detection and the small metadata mismatch are scope notes but not concerning by themselves.
Install Mechanism: okNo install spec or code files are included (instruction-only). Nothing is downloaded or written to disk by an installer as part of the skill package, which reduces installation risk.
Credentials: okOnly one environment variable is declared (NEMO_TOKEN) and it is directly tied to the remote rendering service. The skill instructs creating an anonymous token when none is present — this is consistent with the need for a service credential. The previously noted configPath in SKILL.md is a minor metadata inconsistency but does not indicate overbroad credential access.
Persistence & Privilege: okThe skill is not marked always:true and makes no requests to modify other skills or system-wide settings. It stores a session_id for the rendering session, which is normal for a remote session-based API. Autonomous invocation is allowed by default (platform normal) but not combined with other concerning flags.