Skillv1.0.2

ClawScan security

Wan 2.6 & 2.5 — AI Video & Image Generation by Alibaba · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignMar 18, 2026, 4:38 AM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: The skill's files, env requirements, and instructions are consistent with a video/image generation connector for Atlas Cloud (with optional Google support); nothing in the bundle indicates covert exfiltration or unrelated privileges, though there are a few minor metadata inconsistencies and privacy/billing considerations to be aware of.
Guidance: This skill appears to do what it says: call Atlas Cloud APIs to generate images/videos. Before installing: (1) Be aware ATLASCLOUD_API_KEY gives full model access and billing rights — use an account with spending limits or a separate project key if possible. (2) The scripts will upload local files (with a confirmation prompt) and will send prompts/media to api.atlascloud.ai — don’t upload sensitive files unless you trust the service. (3) The image script can also use Google AI Studio if GEMINI_API_KEY is present, but that env var is not declared in the skill metadata — if you do not want any Google calls, avoid setting GEMINI_API_KEY. (4) If you need more assurance, review the two Python scripts locally (they are small and standard-library only) and test with a low-privilege/budget-controlled API key first.

Review Dimensions

Purpose & Capability: okThe skill is presented as an Atlas Cloud (Wan 2.6/2.5) image & video generation tool and it only requires ATLASCLOUD_API_KEY in metadata. The included scripts call api.atlascloud.ai for model listing, generate, poll, upload, and download operations which is coherent with the stated purpose.
Instruction Scope: noteSKILL.md and the scripts instruct the agent to send prompts, media URLs, audio URLs, and uploaded local files to api.atlascloud.ai. The video script prompts for confirmation before uploading local files. The image script also supports Google AI Studio (generativelanguage.googleapis.com) if GEMINI_API_KEY is present — that provider support is not declared in the skill metadata and is therefore a scope mismatch (optional provider only, not hidden exfiltration).
Install Mechanism: okNo install spec (instruction-only plus two small Python scripts) — nothing is downloaded or extracted from arbitrary URLs. The scripts use only the Python standard library, so the install risk is minimal.
Credentials: noteMetadata declares one env var (ATLASCLOUD_API_KEY), which is appropriate for an Atlas Cloud connector. The code also optionally reads GEMINI_API_KEY for Google AI Studio support but that env var is not declared in requires.env. The SKILL.md warns the Atlas key is not scoped and will be billed under the account — this is important: the key grants model usage and billing authority, so ensure you use an account with appropriate limits/monitoring.
Persistence & Privilege: okThe skill does not request always:true and does not modify system-wide agent settings. It does not persist credentials itself or alter other skills. Autonomous invocation is allowed (platform default) but not combined with other red flags here.