Build Your Own Deep Research Agent + Technical Writer Multi-Agent System
A hands-on workshop repository that builds a two-agent AI system, a deep-research agent and a writing workflow, each served as an MCP server and connected to a coding-agent harness such as Claude Code or Cursor.
This repository accompanies a workshop presented at the AI Engineering Conference Europe that builds a multi-agent AI system out of two MCP servers, a deep-research agent and a LinkedIn writing workflow, both reachable from a harness such as Claude Code or Cursor. It includes a recorded video of the full workshop and its slides.
The repository supports three ways to work through it: watching the roughly two-hour workshop video for the end-to-end mental model of tool-use agents, evaluator-optimizer loops, grounded search, structured output and MCP-server design; running the finished code to see it generate a research brief and a LinkedIn post through an evaluator-optimizer loop, scored by an LLM acting as judge; or implementing the system from a stripped-down skeleton with pre-groomed tickets and a custom Claude Code skill that orchestrates implementer and tester agents in a loop until the rebuilt project matches the reference source, working in an isolated folder that cannot see or read the reference implementation.
It is a two-to-four-hour introduction, and links to a longer paid course covering three end-to-end agent projects for readers who want to go further.