Agent Skills Pack

This repository contains small, focused skill cards that you can load into AI coding assistants (e.g., GitHub Copilot, Cursor, Antigravity, Replit Agents). Each skill is a short markdown file that teaches the assistant conventions and patterns for a given topic.

Contents

• Brainstorming (ideation prompts)
• LangChain (LLM orchestration, tools, RAG)
• LangGraph (stateful graphs, agents, streaming)
• React best practices
• Web design guidelines

How to use with your AI assistant

1. Open or import skills

• Open the relevant skills/<topic>/SKILL.md in your IDE so the assistant can read it.
• If your assistant supports a “knowledge” or “context” panel, pin the SKILL file there.

2. Tell the assistant to use the skill

• Add a brief instruction in your prompt, e.g., “Use the LangGraph skill from SKILL.md while editing this file.”
• When switching tasks, mention the specific skill to apply.

3. Keep scope small

• Ask for one change at a time (e.g., “add a LangChain RetrievalQA chain here”).
• Prefer concrete goals and files; avoid multi-file broad requests when possible.

4. Let the assistant cite lines

• When you ask for edits, request line-linked references so you can jump to changes quickly.

5. Refresh when files change

• If you update a SKILL file, remind the assistant to re-read it before continuing work.

Suggested prompts

• “Use the LangChain SKILL.md to add a tool-enabled agent.”
• “Follow the Web Design SKILL.md while refining the landing page styles.”
• “Apply the React best practices SKILL.md to refactor this component.”
• “Reference the Brainstorming SKILL.md to propose feature ideas.”

Contributing

• Keep skills concise and technology-specific.
• Favor runnable snippets over long prose.
• Use ASCII and add minimal comments only where code is non-obvious.
• Add new skills where useful and open a PR with a short note on what the skill teaches.