Sources & further reading
This course is a plain-language synthesis. Below is where the claims come from, and what to read next if you want the primary material.
The content was distilled from a curated research vault of 84 notes organized across twelve areas: foundations, architectures, frameworks, context engineering, model architectures, state management, tool use, multi-agent systems, domains, evaluation, production and frontiers. Each module below maps to the notes it draws on, so the deeper material stays available for the reader who wants it.
C.1Module → source notes
| Module | Source notes |
|---|---|
| 01 · What an agent is | Foundations/Agent-Definitions.md, Foundations/Core-Abstractions.md, Architectures/Human-in-the-Loop.md |
| 02 · How the model thinks | Foundations/LLM-Fundamentals.md, ModelArchitectures/Reasoning-Models.md, ModelArchitectures/Inference-Optimization.md |
| 03 · The loop | Harness/Fundamentals/Agent-Loop.md, Harness/Fundamentals/Harness-Architecture.md, Harness/Fundamentals/Design-Principles.md |
| 04 · Tools | ToolUse/Function-Calling.md, ToolUse/MCP-Protocol.md, Harness/Core-Components/Permission-System.md, ToolUse/Code-Execution.md |
| 05 · Context & memory | ContextEngineering/Context-Window-Management.md, ContextEngineering/RAG-Fundamentals.md, ContextEngineering/Advanced-RAG.md, ContextEngineering/Memory-Systems.md |
| 06 · Patterns | Architectures/Single-Agent-Patterns.md, Architectures/Control-Flow.md, ModelArchitectures/Model-Routing.md |
| 07 · Many agents | MultiAgent/Coordination-Patterns.md, MultiAgent/Delegation-Handoff.md, Harness/Execution/Sub-Agents.md |
| 08 · Evaluation | Evaluation/Agent-Benchmarks.md, Evaluation/Metrics.md, Evaluation/Red-Teaming.md, Evaluation/Offline-Eval.md |
| 09 · Production | Production/Cost-Optimization.md, Production/Observability.md, Production/Reliability.md, Production/Security.md |
| 10 · Start here | Frameworks/Framework-Comparison.md, Harness/Building-A-Harness/Minimal-Harness.md, Harness/Building-A-Harness/Reference-Implementations.md |
C.2The primary papers
Six papers cover most of what this course describes. They are readable with patience, and reading one properly is worth ten summaries.
| Paper | What it introduced | Where |
|---|---|---|
| Attention Is All You Need 2017 | The transformer — the architecture under every model in this course | arXiv:1706.03762 |
| Language Models are Few-Shot Learners 2020 | GPT-3; scale plus in-context learning | arXiv:2005.14165 |
| Chain-of-Thought Prompting 2022 | Reasoning as explicit intermediate steps | arXiv:2201.11903 |
| ReAct: Synergizing Reasoning and Acting 2022 | The think → act → observe loop this course is built around | arXiv:2210.03629 |
| SWE-bench 2023 | Real repository issues as an agent benchmark | arXiv:2310.06770 |
| GAIA: A Benchmark for General AI Assistants 2023 | Multi-step questions requiring tools and browsing | arXiv:2311.12983 |
C.3Documentation worth bookmarking
C.4License and reuse
This course is released under CC BY 4.0: copy it, adapt it, teach from it, including commercially — with credit to Starphix and a link to the license.
The prose, diagrams and exercises are original synthesis built on public research and documentation. The papers, standards and product names remain the property of their authors and are cited rather than reproduced. Prices, model names and benchmark figures date quickly; check them against a current source before quoting them in a classroom or a proposal.
v1.0