Readings.
Start Here — if you read nothing else, read these
Andrew Ng's practical advice on growing with AI. Written for everyone, not just engineers. 10 min read.
The format that makes AI 10x more efficient. Learn the basics in 5 minutes. You'll use this every day.
What Claude can do today. Written for a general audience. Understand the tool before you use it.
Prompting & Working With AI
Official guide on writing better prompts for Claude. Covers structure, clarity, and common patterns.
What "vibe coding" really means. A balanced take on using AI to write code when you're not a developer.
Why token efficiency matters. What happens when AI "forgets" things. How to keep conversations productive.
A Wharton professor's real-world advice on using AI at work. Written for non-technical professionals.
GitHub & Knowledge Management
GitHub's official beginner tutorial. Create a repo, make a branch, commit a change. Takes 10 minutes.
How to edit files directly in the browser. No terminal needed. This is how you'll update your CLAUDE.md.
How to create, manage, and revoke GitHub tokens. Essential for connecting GitHub to Claude.
Shared Craft & AI-SDLC — S6 reads (23 Apr 2026)
The ongoing Thoughtworks memo series on building with AI in production. Aditya’s “rework tax” and the pipeline response both show up repeatedly. If S6’s thesis landed, this is the long-form version of it.
The 150M-lines-of-code study behind the 41% revert-rate stat in S6. Churn, copy-paste and duplication metrics on AI-assisted code. Bring this to any internal pitch on why the pipeline matters.
Layer 2 of Aditya’s pipeline lives here. Pre-commit, pre-push, post-session — the same primitives Safe-to-Ship will ride on. Skip S4? This is the catch-up.
Why AI gets you 70% of the way there and why the last 30% is the whole job — the same argument Aditya made about demo-ware vs. production. Short, sharp, and quotable.
Context & Session Management — S5 bonus read (16 Apr 2026)
Published the same day as Session 5. Practical guide to the five things you can do after any Claude Code turn — continue, /rewind, /clear, /compact, or spawn a subagent. With 1M context, session hygiene matters more, not less. Includes the rule-of-thumb for when to start a fresh session versus compacting in place.
Context rot, compaction, and the 1M window. Short. Authoritative. Read alongside Thariq’s practical version.
Skills, Plugins, MCPs & Agents — S5 core reads
The technical rationale and architecture behind Agent Skills. Three-tier loading, progressive disclosure, and why it changes token economics at org scale. Start here before reading anything else on this topic.
Practical decision framework for choosing between skills, workflows, and agents. Cuts through the confusion with concrete examples. Directly relevant to S5.
Anthropic’s official course on Agent Skills. Covers the three tiers, how to write skills, and enterprise management. Free. Directly from the team that built the standard.
The open standard for connecting AI to your tools. Adopted by Anthropic, Microsoft, OpenAI, GitHub, and Cursor. Understand what an MCP is before using one.
Claude Code & CLI — the full-power workflow
Official overview of Claude Code. What it is, how it works, what it can do that Claude UI cannot. Start here before Session 4.
Pre-hooks and post-hooks explained. How to automate guardrails that check your work before and after every action.
How to connect your personal tools (Slack, Gmail, Calendar, Linear) directly to AI. The future of AI-tool integration.
How Claude Code uses CLAUDE.md and memory files. The knowledge architecture behind persistent AI context.
Security & Best Practices — guardrails for AI-assisted work
Official security model for Claude Code. Permission modes, sandbox, and how to keep AI-generated code safe.
How GitHub detects leaked secrets (API keys, tokens) in your code. Essential when AI is writing commits.
How to review AI-generated code. What to look for, what to trust, what to verify. Written by a practitioner.
Courses — go deeper when you're ready
Free course designed for non-technical people. Covers what AI can do, how to spot opportunities, and how to work with AI teams. 6 hours total.
Short course on writing effective prompts. Developer-oriented title but the principles apply to everyone. 1 hour.
Interactive tutorials to learn GitHub. Start with "Introduction to GitHub" — it walks you through everything in your browser.
Stay Current — newsletters worth reading
Weekly AI newsletter. Practical, accessible, no hype. The best way to stay current without drowning in Twitter.
A professor who actually uses AI daily. Practical insights on AI at work. Written for normal people.
Official updates from the team behind Claude. New features, capabilities, and research — straight from the source.