Skillv1.0.1

ClawScan security

magic-image2video · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignMar 17, 2026, 2:51 AM

Verdict: Benign
Confidence: medium
Model: gpt-5-mini
Summary: The skill's code, instructions, and env requirements consistently match its stated purpose: it uses a Python client and a single MAGIC_API_KEY to submit text+images to a remote video API and poll for the resulting video link.
Guidance: This skill appears to be what it says: it will use your MAGIC_API_KEY and the bundled Python script to submit text and images to https://open-test.magiclight.ai, upload any local image files you provide, and poll for the resulting video link. Before installing or invoking it: - Only provide images you are comfortable having uploaded to a third-party service (local files will be transmitted). - Confirm you trust the endpoint (open-test.magiclight.ai) and that your MAGIC_API_KEY is intended for use with that service; treat the key like a password and rotate it if you later suspect misuse. - Provide only public image URLs; avoid giving internal or localhost URLs (the client will fetch them), to reduce SSRF/exfiltration risk. - Ensure the Python runtime has required packages (certifi is imported) or run in an isolated/test environment first. - If you do not trust the skill source or endpoint, do not supply sensitive images or credentials; consider running the script manually in a controlled environment to observe behavior.

Review Dimensions

Purpose & Capability: okName/description ask for text+image→video submission; the included Python script calls a remote API (BASE_URL https://open-test.magiclight.ai) to create tasks, upload local images, and poll for results. Required binary (python) and env var (MAGIC_API_KEY) are appropriate for this functionality.
Instruction Scope: noteSKILL.md instructs the agent to extract TEXT and IMAGE from the user, run the provided script's video-create/video-wait commands, parse JSON stdout, and report task_id/video_url. This stays within the stated purpose. Important note: the skill explicitly supports local file paths and will upload local image files to the remote service — that means local files may be transmitted off-host, which is expected behavior but privacy-relevant.
Install Mechanism: okThere is no install spec (instruction-only with a bundled script). No remote downloads or archive extraction occur. One minor inconsistency: the script imports 'certifi' (a Python package) but the skill does not declare Python package dependencies; ensure the runtime environment has certifi (or install it) before running.
Credentials: okThe only required environment variable is MAGIC_API_KEY, which the client uses in an Authorization header to contact the remote service — this is proportionate. No unrelated credentials or extra environment access are requested. Reminder: providing the API key grants the skill ability to act on behalf of that key at the configured BASE_URL.
Persistence & Privilege: okThe skill does not request always:true, does not modify other skills or system config, and is user-invocable. It does network I/O and file uploads, but no elevated or persistent platform privileges are requested.