Reading and reference library
These links are copied from the supplied course direction (draft of 4 November 2026). Stanford CS329Z and the Google/Kaggle series are references for this course; they are not identified as its provider. Publication and availability statements in the class plans describe the draft, not a fresh availability check.
Part 1 starting points
- Classes 1–2: Stanford CS329Z Lectures 1–2, Compound AI Systems, Ollama, Tokenizers, and Tam et al.
- Class 3: CS329Z Lectures 1 and 4, ReAct, and Demystifying evals for AI agents.
- Use the paired lab's required-materials list for staff datasets, skeletons, and answer keys.
Source library
- Stanford CS329Z — Lecture 1 slides
- Stanford CS329Z — Lecture 2 slides
- Zaharia et al. — Compound AI Systems
- Ollama documentation
- Hugging Face Tokenizers
- Tam et al. — Let Me Speak Freely?
- Stanford CS329Z — Lecture 4 slides
- Stanford CS329Z schedule
- Anthropic — Demystifying evals for AI agents
- Yao et al. — ReAct
- Stanford CS329Z — Lecture 3 slides
- Lewis et al. — Retrieval-Augmented Generation
- rank_bm25
- Sentence Transformers
- Google/Kaggle — Agent tools and interoperability
- OWASP Top 10 for Agentic Applications 2026
- MCP specification
- Schick et al. — Toolformer
- MCP Python SDK
- Google/Kaggle — The new sdlc with vibe coding
- Stanford CS329Z — Lecture 6 slides
- Stanford CS329Z — Lecture 7 slides
- Lindenbauer et al. — The Complexity Trap
- Mem0 paper
- Google/Kaggle — Agent skills
- Claude Code — Hooks
- Claude Code — Permissions
- Claude Code — Settings
- Claude Code — Memory
- Böckeler — Harness engineering
- Claude Code — Skills
- Google/Kaggle — Spec driven production grade development in the age of vibe coding
- Google/Kaggle — Vibe coding agent security and evaluation
- Böckeler — TDD in the agent loop
- Kiro specs
- Claude Code — Sandboxing
- Greshake et al. — Indirect Prompt Injection
- Stanford CS329Z — Lecture 10 slides
- Stanford CS329Z — Lecture 11 slides
- Zhu et al. — Rigorous Agentic Benchmarks
- Stanford CS329Z — Lecture 5 slides
- Stanford CS329Z — Lecture 8 slides
- Arena — HarnessTax
- Addy Osmani — Agentic Autonomy Levels
- Yang et al. — SWE-agent
- Claude Code — Headless
- Kapoor et al. — AI Agents That Matter
- Stanford CS329Z reference course
Original direction
Download the full 65-page PDF. It also preserves the class-priority scoring rubric, instructor planning notes, review rationale, discarded options, and open decisions. Those rankings are curriculum planning scores, not student grades.