Skillv1.0.4

ClawScan security

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

Scanner verdict

BenignMar 18, 2026, 6:27 AM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: The skill's requests and runtime instructions are consistent with a command-line faceswap client that calls verging.ai; required binaries and the single API key requested are proportionate to the stated purpose, though uploading sensitive videos/faces to a third‑party service and the use of public CDN URLs have privacy implications the user should consider.
Guidance: This skill appears to do what it says: it downloads remote videos (when requested), optionally trims them, uploads video and face images to verging.ai via presigned URLs, and creates a face-swap job using your VERGING_API_KEY. Before installing or using it, consider: 1) Privacy: uploading videos and face images to a cloud service can expose sensitive content — the sample response shows a public_url on a different domain (img.panpan8.com), so check verging.ai’s privacy/retention rules and whether results are publicly accessible. 2) Billing: face-swap jobs consume credits (HD costs more) — confirm cost and quota on your account. 3) API key safety: treat VERGING_API_KEY like a secret, and revoke it if you suspect misuse. 4) Legal/ethical: ensure you have rights to reuse source videos and faces. 5) Dependencies: the skill relies on external binaries (yt-dlp, ffmpeg); installing community skills via npx will fetch code from npm — only run those install commands if you trust the source.

Review Dimensions

Purpose & Capability: okName/description match what the skill requires: yt-dlp for remote downloads, ffmpeg/ffprobe for trimming, and curl plus VERGING_API_KEY to call verging.ai. Declared permissions (network, filesystem, external_commands) are expected for this functionality.
Instruction Scope: noteInstructions are narrowly focused on downloading (yt-dlp/curl), trimming (ffmpeg/ffprobe), uploading to presigned URLs, and polling verging.ai job endpoints. Two points to note: the example API response includes a public_url on img.panpan8.com (a domain different from verging.ai), which implies uploaded media may be hosted publicly or on a third-party CDN; and temp files are written to /tmp/verging-faceswap. There are no instructions to read unrelated system files or additional environment variables.
Install Mechanism: okThis is instruction-only (no install spec), which is low risk. The skill suggests installing other community skills (yt-dlp) via npx if needed; those are external actions the user would perform, not automatic installs by this skill.
Credentials: okThe only required credential is VERGING_API_KEY (primaryEnv) and an optional VERGING_API_URL override — these directly match the described API usage. No unrelated secrets or system credentials are requested.
Persistence & Privilege: notealways:false (no forced inclusion). The skill's required permissions (network, filesystem, external commands) are necessary for operation, but combined with an API key they allow the agent to perform remote uploads and start jobs. This is expected for a remote faceswap service; users should be aware the agent can initiate network activity using their API key when invoked.