A modular architecture of 84 production-tested domain skills for Claude Code and Antigravity IDE. Covers AI Agents, RAG, Software Engineering (TDD, Security), PM frameworks, and Automation. Includes a free 5-skill open source starter kit.
Hey Product Hunt! š
I'm Yevhen, an AI Product Manager & Full-Stack AI Engineer.
Like many developers using Claude Code and agentic coding tools daily, I kept hitting "context decay" ā where monolithic system prompts bloat the context window and degrade reasoning quality over long sessions.
To solve this, I built the Claude AI Skills Pack: 84 modular, production-tested domain skills organized into 7 categories (AI Agents, RAG & Vector Search, Software Engineering, Product Management, Automation, Content, and Growth).
Each skill is a structured SKILL.md file that Claude automatically indexes and activates only when you work in that specific domain.
š What's inside the full suite:
- š¤ 17 AI Agent & Multi-Agent systems (LangGraph, MCP servers, Fleet governance)
- š 5 RAG & pgvector hybrid search skills
- āļø 21 Engineering skills (TDD workflows, Security reviews, API contracts)
- š 14 Product Management skills (PRD generators, Prioritization frameworks)
- š§ 27 Data, Automation & Business skills
We also open-sourced a free 5-skill starter kit on GitHub so you can test the architecture right away: https://github.com/yevhens-hue/c...
I'd love to hear your feedback, how you currently handle agentic guardrails, and answer any questions!