Skip to content

Chapters

This textbook is organized into 20 chapters covering 379 concepts.

Chapter Overview

  1. Research Workflows and the Researcher as Orchestrator - Introduces the research workflow, the five everyday research tasks, core generative AI vocabulary, and the central idea of the researcher as an orchestrator of agent teams.
  2. How Language Models Work - Explains tokens, context windows, prompts, training data, and hallucination so researchers understand why models produce fluent but sometimes false output.
  3. Choosing Models and Writing Clear Prompts - Covers choosing among models by size, cost, and access, adding retrieval, and writing clear, constrained prompts with examples and output formats.
  4. Context Engineering and Advanced Prompting - Develops context engineering, grounding, prompt testing, and multi-step prompting, along with defenses against prompt injection and long-document problems.
  5. Markdown Project Files and Version Control - Teaches Markdown, project instruction files such as AGENTS.md and CLAUDE.md, and the Git practices that keep project knowledge organized and reliable.
  6. Knowledge Bases, Personal Knowledge Graphs, and the Anatomy of an Agent Skill - Builds project knowledge through notes, logs, and knowledge bases, turns a researcher's notes into a personal knowledge graph that agents can query, then introduces the parts of an agent skill.
  7. Designing, Testing, and Sharing Agent Skills - Shows how to scope, test, version, share, and maintain agent skills, ending with a template for turning a repeatable research task into a skill.
  8. Agents, Tool Use, and Subagents - Describes the agent loop, tool calls, file and shell access, sandboxing, permissions, the Model Context Protocol, and subagents that work in parallel.
  9. Supervising Agents and Defining Agent Roles - Covers autonomy, supervision, logging, failure modes, and cost control for agents, then defines the planning, execution, and reviewer roles agent teams need.
  10. Orchestrating Teams of Planning and Execution Agents - Shows how to coordinate teams of planning and execution agents through task queues, handoffs, approval gates, delegation, and scaling decisions.
  11. AI-Assisted Literature Search and Synthesis - Applies agents to search strategy, paper triage, summarization, citation verification, and evidence synthesis for rigorous literature reviews.
  12. AI-Assisted Data Analysis and Scripting - Uses AI to clean, explore, visualize, and analyze measurement data with Python scripts and notebooks while keeping errors visible.
  13. Simulation, Testing, and Reproducible Code - Covers unit tests, debugging, silent errors, simulation and HDL code, parameter sweeps, and reproducible environments for AI-written research code.
  14. Brainstorming and Problem Framing - Applies AI to framing problems, generating and ranking ideas, developing hypotheses, designing experiments, and checking novelty.
  15. AI-Assisted Manuscript Preparation - Walks through paper structure, drafting, revision, author voice, LaTeX, figures, references, peer review responses, and arXiv submission with AI assistance.
  16. MicroSims and Research Communication - Uses slide decks, interactive simulations, MicroSims, learning graphs, and intelligent textbooks to communicate and teach research results.
  17. Verification and Reproducibility - Presents verification and validation methods, including independent and cross-model checks, replication, audit trails, and tracing errors to their source.
  18. Responsible and Ethical Use of AI - Addresses citation and disclosure of AI use, privacy, intellectual property, bias, institutional and journal policy, accountability, and over-reliance.
  19. Always-On Personal Agents for Research - Surveys always-on personal agents such as Grok Bot, Dots, Muse, Claude Cowork, OpenClaw, and Hermes Agent, then shows how to compare, secure, and select them and how a personal knowledge graph serves as their memory.
  20. Case Studies and Capstone Project - Applies the whole toolkit through case studies, including the MicroSims paper and intelligent textbook project, and closes with the capstone project and adoption plan.

How to Use This Textbook

Work through the chapters in order. Each chapter lists the earlier chapters it builds on, and every concept appears in exactly one chapter, after the concepts it depends on. Early chapters build the foundations of models, prompts, project files, and skills; middle chapters apply agents to the five research tasks; late chapters cover verification, ethics, always-on personal agents, and the capstone.

Note: Each chapter includes a list of concepts covered. Make sure to complete prerequisites before moving to advanced chapters.