Impeccable is triggered when designing, redoing, reviewing, or polishing frontend interfaces, covering websites, landing pages, dashboards, forms, empty states, as well as native iOS and Android interfaces. It can also organize DESIGN.md files and design tokens. At the start of each session, it runs its built-in context command to read product and design specifications, then loads corresponding practices based on subcommands. Before any interface changes, it must read the 'Craftsmanship Baseline' file and view the rendered results at actual size.
Why It Matters It incorporates hierarchy, typography, color schemes, animations, accessibility, and edge states into its checks, requiring review of actual rendered results before delivery to reduce a template-like feel.
skills.sh installs 406503 (+2352 / 24h)
Security Audit: Risk warnings present
- Installs or runs third-party packages (npx / pip, etc.), which constitutes executing commands
- References unscanned scripts: scripts/impeccable, scripts/impeccable.cmd
- Audit scope: GitHub pbakaus/impeccable/.agent/skills/impeccable/SKILL.md (only SKILL.md text was scanned)
Rationale
- skills.sh 24 hour installs 2352 (indicates attention only); Apache 2.0 license
- The unscanned scripts flagged in the audit are its built-in impeccable launchers; SKILL.md documents fallbacks if the launcher fails
Usage
- In a frontend project, ask the agent to 'redesign this settings page' or 'review this landing page'; it will first read the product and design specifications within the project
Risks
- The launcher downloads a standalone binary file on first run, the contents of which are outside the scope of this scan