ai-coding-starter-kit
Kuratierte Agent Skills, Checklisten, Templates und Leitfäden für Schweizer Entwicklungsteams – direkt aus meinen Blog-Artikeln destilliert.
This repository is a curated collection of resources aimed at developers who want to start using AI coding assistants in their daily work. The README is written in German and the project is associated with a Swiss developer blog called agentic-coding.ch. It was assembled by Dr. Rene Bader and is released under the GNU General Public License v3.0.
The kit focuses on four AI coding tools: Claude Code, OpenAI Codex, Cursor, and Gemini CLI. Rather than being a piece of software itself, it acts as a reference library, pointing developers to pre-built add-ons called agent skills, along with configuration file templates and security review checklists. Each skill is a small package that extends what an AI assistant can do, for example web scraping, structured planning, UI design guidance, or automated code security audits.
The skills section lists about eleven curated options, each with a brief description of what it handles and a single install command. Examples include a planning skill that helps the AI track multi-step tasks across files, a Vercel design guidelines skill for interface and accessibility work, and a Trail of Bits security audit skill that runs tools like CodeQL and Semgrep. The templates section covers ready-made configuration files for Claude, OpenAI, Gemini, and Gemma, including several focused on code review, full-stack development, and security testing scenarios.
A notable part of the kit is its emphasis on caution. The README and an accompanying SECURITY.md file walk through a checklist developers should follow before installing any skill: reviewing the source, checking the license, inspecting install scripts and dependencies, and testing in an isolated environment before using it in production. The project treats AI agent skills the same way a team would treat a third-party developer tool, something to evaluate carefully rather than install blindly.
The project is aimed at teams that want a structured, safety-conscious entry point into AI-assisted development workflows, covering the full cycle of planning, coding, testing, reviewing, and securing software.