Skillv1.0.0

ClawScan security

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

Scanner verdict

BenignApr 23, 2026, 3:08 AM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: The skill's requirements and runtime instructions are coherent with its stated purpose (uploading media and driving a cloud video-rendering API); nothing requested is disproportionate or unrelated to creating product videos.
Guidance: This skill sends any media you provide to mega-api-prod.nemovideo.ai for cloud rendering and will use either your NEMO_TOKEN (if present) or obtain a short-lived anonymous token. Before installing or using it, confirm you are comfortable uploading the images/video (no sensitive content), and avoid pasting long-lived secrets unless you trust the provider. Note the small inconsistency: the registry marks NEMO_TOKEN as required, but the runtime flow can obtain an anonymous token if none is present. If you plan to provide your own NEMO_TOKEN, ensure it is scoped appropriately. Finally, review the service’s privacy/terms (where available) and be aware the agent may check a couple of local paths to set an attribution header — this only probes known install locations and is used for telemetry/attribution according to the instructions.

Review Dimensions

Purpose & Capability: okThe skill is a cloud-backed product-video creation helper and only asks the agent to talk to nemovideo.ai and to use a single service token (NEMO_TOKEN). The declared metadata (NEMO_TOKEN, config path) aligns with interacting with a backend video service and with uploading/downloading media.
Instruction Scope: noteInstructions describe creating sessions, uploading user-provided media, posting SSE messages, polling render status, and returning download URLs — all within the expected scope. Two small points to note: (1) the agent is instructed to detect an install path (~/.clawhub, ~/.cursor) to set an attribution header (this requires reading the local filesystem), and (2) uploads reference local file paths for multipart posts (expected for uploading user files). Both are consistent with the skill's purpose but mean the agent will access files the user supplies and may probe a couple of well-known paths.
Install Mechanism: okThere is no install spec (instruction-only), so nothing is written to disk by the skill itself. This is the lowest-risk install model.
Credentials: noteThe only declared credential is NEMO_TOKEN, which is proportionate for a cloud API. However, the SKILL.md includes a fallback anonymous-token flow when NEMO_TOKEN is absent, so the registry's 'required env var' claim is stricter than the runtime instructions; this is a mild inconsistency but not a security red flag. The metadata also references a config path (~/.config/nemovideo/) which is plausible but not strictly necessary based on the instructions.
Persistence & Privilege: okThe skill does not request always:true and is instruction-only. It does not modify other skills or system-wide settings. Autonomous invocation is allowed (platform default) but not combined with any broad or unrelated credential access.