Resources

Pre-event reading and setup. The Monday tutorial assumes nothing beyond Python, but skimming the first two reads beforehand will make the morning go much faster.

Start here

• Building Effective Agents Anthropic

  The canonical short read on what an agent actually is — workflows vs. agents, the common patterns, and when not to build one. Start here if you're new to this.

• Harness, Scaffold, and the AI Agent Terms Worth Getting Right Hugging Face

  A quick glossary if the terminology is unfamiliar — model vs. harness vs. scaffold, tools, skills, sub-agents.

Set up before you arrive

• Claude Code Quickstart Anthropic

  Install Claude Code and run a first task. We'll use it in the Monday tutorial — having it working saves you the setup scramble.

• github.com/agents4academia GitHub

  The hackathon org. We expect every team to maintain their project repository here.

Building blocks

• LiteLLM docs

  One Python interface to 100+ LLM providers. The simplest way to make direct model calls — where the tutorial starts.

• Claude Agent SDK overview Anthropic

  For when you want to write your own agent rather than drive an off-the-shelf one.

• Model Context Protocol modelcontextprotocol.io

  What MCP is and how agents connect to external tools and data. See also the ready-made servers you can plug in — filesystems, GitHub, databases, and more.

Going deeper

• Effective Context Engineering for AI Agents Anthropic

  Managing what the model sees at each step — the skill that most separates working agents from flaky ones.

• Agent patterns cookbook Anthropic

  Runnable notebook implementations of the patterns from Building Effective Agents — prompt chaining, routing, parallelization, orchestrator–workers.