Security audit
Security checks for vulnerabilities and agentic risk
This design skill is a coherent guidance package with disclosed, limited preference storage and no executable or hidden high-impact behavior.
Before installing, be aware that the skill can reuse and update design preferences under ~/Clawic/data/design/. Review those files if you do not want design choices or recurring feedback remembered across sessions.
Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.
last: YYYY-MM-DD
## Context
<!-- Their projects and typical artifacts (product UI, decks, social, print) -->
<!-- Tools and delivery format (HTML/CSS, Figma specs, rendered images) -->
## Brand and Register
Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.
| State | Rule |
|---|---|
| Empty | Teach, don't apologize: what this area will show + the action that fills it. Never a bare "No data" |
| Loading | Skeleton in the final layout (prevents shift); spinner only for sub-second or indeterminate whole-screen waits |
| Error | What failed, in user terms + retry action. Keep the layout; don't collapse to a blank error page for a partial failure |
| Overflow | Decide truncation per element: ellipsis + tooltip for identifiers, line-clamp for descriptions, "+N more" for chips. Test with 3x expected content length and German-length words |
The skill’s activation description is extremely broad and matches many generic design-related requests such as 'looks off,' choosing fonts or colors, adapting mobile or dark mode, and reviewing generated HTML/CSS. This can cause the skill to trigger in situations where a more specific or safer skill should apply, increasing the chance of misrouting, over-collection of user context, or unintended file/memory interactions under this skill’s scope.
The skill instructs the agent to read from and write to persistent files under the user's home directory, including config and memory files, without any user-facing disclosure or explicit consent step. This creates a privacy and integrity risk because the agent may silently collect, persist, and reuse user preferences or project context across sessions, and may overwrite stored data based on conversational statements the user did not realize would be memorialized.
No suspicious patterns detected.