iusztinpaulCommunity·22 Agent Skills Included

designing-real-world-ai-agents-workshop

Hands-on workshop building research and writing MCP servers

Teaches you to build a deep research system and a LinkedIn writing workflow as MCP servers connected to Claude Code or Cursor. Covers grounded search, structured output, evaluator-optimizer loops, and LLM-as-judge evaluation with working Python code. Includes 25 pre-groomed tickets and guided implementation loops so you rebuild the full system step by step.

487|128

↗

npx skills add iusztinpaul/designing-real-world-ai-agents-workshop --all -g -y

Available:

Tells your AI agent the project structure, tech stack, and how to run, test, and invoke the research and writing MCP servers through Make targets, server processes, or slash-command skills.

All Skills in This Repository (22)

Pure Emerald Level Indicators

📦 In Repo

iusztinpaul

developing-with-streamlit

Route Streamlit development tasks to targeted sub-skills for building, debugging, and theming.

using-streamlit-layouts

Organize Streamlit app content into sidebars, columns, containers, and dialogs.

organizing-streamlit-code

Enforce modular directory layouts for Streamlit apps with separated UI and logic.

using-streamlit-markdown

Render GitHub-flavored Markdown and Streamlit extensions in st.markdown workflows.

using-streamlit-cli

Automates Streamlit CLI commands for running, configuring, and diagnosing apps.

optimizing-streamlit-performance

Optimize Streamlit apps using caching APIs and UI fragmentation to reduce reruns.

building-streamlit-dashboards

Generate interactive Streamlit dashboards with KPI cards, sparklines, and responsive layouts.

improving-streamlit-design

Improves Streamlit app visuals with Material icons, badges, spacing, and typography rules for dashboards and data apps.

using-streamlit-custom-components

Install and integrate third-party custom components into Streamlit apps.

connecting-streamlit-to-snowflake

Connect Streamlit apps to Snowflake using st.connection and parameterized queries.

using-streamlit-session-state

Persist and manage UI state across Streamlit reruns using st.session_state.

building-streamlit-custom-components-v2

Create Streamlit custom components with React/TypeScript and state syncing.

Community

Advanced

Frequently Asked Questions

FAQPage Schema

How to install designing-real-world-ai-agents-workshop?▼

Run `npx skills add iusztinpaul/designing-real-world-ai-agents-workshop --all -g -y` in your terminal to install all skills in this suite globally.

What will I build in this workshop?▼

You build two MCP servers: a deep research system using Gemini with Google Search grounding, and a LinkedIn post writer with an evaluator-optimizer review loop.

Does this workshop work with Claude Code and Cursor?▼

Yes. Both MCP servers follow the Model Context Protocol standard and connect to Claude Code, Cursor, or any MCP-compatible harness.

Can beginners follow this workshop?▼

Yes. You can watch the 2-hour video, run the finished code in 30 minutes, or rebuild everything yourself using the 25 guided tickets in the implement_yourself folder.

What AI engineering patterns does it teach?▼

It covers tool-use agents, evaluator-optimizer loops, grounded search, structured LLM output with Pydantic, MCP server design, and LLM-as-judge evaluation with Opik.