Back to skill
Skillv1.0.1

ClawScan security

Huo15 Openclaw Brand Protocol · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignApr 24, 2026, 4:16 AM
Verdict
benign
Confidence
high
Model
gpt-5-mini
Summary
The skill's requested actions and outputs align with its stated purpose (fetching and codifying brand visual specs); it is instruction-only, asks for no credentials, and returns CLI steps for the user to run rather than executing anything itself.
Guidance
This skill is instruction-only and internally coherent, but before using it: (1) verify you trust the skill source and the target URLs you provide; the skill will prompt you to run curl/npx/ImageMagick/grep commands — inspect those commands before running them and run them in a safe account or sandbox if unsure; (2) downloaded assets may be copyrighted or require permission — ensure you have the right to download/use them; (3) the agent will not execute commands itself or ask for credentials, so any network requests or file writes only happen if you run the suggested commands; (4) expect brittle heuristics (SVG fills, CSS grep) and manually review results — the skill provides a verification step and confidence rating for this reason.

Review Dimensions

Purpose & Capability
okName and description (抓取品牌视觉规范并产出 brand-spec.md) match the SKILL.md instructions. The skill does not request unrelated binaries, credentials, or config paths and only prescribes how to find, download, verify, and codify brand assets — which is proportionate to its purpose.
Instruction Scope
noteThe SKILL.md gives a tight 5-step workflow and explicitly avoids running child processes itself, instead returning CLI commands (curl, npx playwright-core, ImageMagick examples, grep) for the user to run. This is coherent with the stated constraint but means the user (not the agent) will perform network downloads and local file inspection; the document references only user-supplied URLs and local ~/brand-kits paths, not other system secrets. Users should be cautious about running curl/npx commands against untrusted URLs.
Install Mechanism
okNo install spec and no code files — instruction-only. Nothing is downloaded or written by the skill itself, reducing installation risk.
Credentials
okThe skill declares no required environment variables, credentials, or special config paths. The SKILL.md does not instruct reading credentials or unrelated system files.
Persistence & Privilege
okalways is false and the skill is user-invocable; it does not request persistent presence or modification of other skills or system-wide settings. It suggests storing downloaded assets under the user's ~/brand-kits directory if the user runs the provided commands.