Petar

Curated skills and commands for AI coding agents. Works across Claude Code, OpenCode, Codex, and Amp.

Browser automation with persistent page state. Navigate websites, fill forms, take screenshots, extract data, and test web apps. Supports standalone Chromium or connecting to your existing Chrome session.

Create distinctive, production-grade frontend interfaces that avoid generic AI aesthetics. Build web components, landing pages, dashboards, and applications with exceptional design quality.

Analyze web performance using Chrome DevTools MCP. Measure Core Web Vitals (FCP, LCP, TBT, CLS), identify render-blocking resources, and optimize page load performance.

Graph-based issue tracker that survives conversation compaction. Track complex multi-session work with dependencies, priorities, and audit trails persisted in git.

Create a SHIP.md document - a concrete plan for finishing a project. Analyzes the codebase, defines done, and breaks work into phases with small actionable steps.

Generate concise, actionable implementation plans for coding tasks. Outputs a structured checklist with scope, action items, and open questions.

Generate structured learning guides using the Polya problem-solving method. Creates self-contained Markdown documents that guide users through building features themselves.

Interactive teaching mode that guides users to complete tasks themselves through Socratic questioning and graduated hints, rather than doing the work for them.

Guide for creating effective agent skills. Use when building new skills that extend Claude's capabilities with specialized knowledge, workflows, or tool integrations.

Clarify requirements before implementing. Ask focused questions to avoid wrong work when requirements are ambiguous.