AgentsCurating

AI Agents & Memory

Reasoning, tools, memory, and long-horizon behavior

A course track for moving from agent primitives to observable, memory-aware systems.

Learning map

This track treats an agent as a system: model decisions, tools, state, memory, evaluation, and the surrounding runtime all matter.

Resource index

6 links

A hands-on course covering agent theory, design, frameworks, use cases, and evaluation.

The official units combine explanations, exercises, and practical material.

Local reading and implementation materials for ReAct, RAG, MCP, and multi-agent systems.

Working notes for memory mechanisms and context management.

Follow-up writing on agent memory, context, and evaluation.

Course resources and published material on the learning site.

DataHub workflow

Agent fundamentals

Tools and orchestration

Memory and retrieval

Evaluation and observability

Applied case studies

Syllabus

Messages, thoughts, actions, observations, tools, and the loop that connects them.

Short-term state, long-term memory, retrieval, compression, updating, and conflict handling.

Tracing, failure analysis, benchmark design, and the feedback loops needed for reliable agents.