Skip to main content
Keep these open beside Claude. Each row is a living guide, PDF, or ebook-style reference. Preview on the left, open on the right.

​
Claude product and prompting

Official guide for clarity, examples, thinking, tools, and agentic prompts across current Claude models.Open How system prompts, skills, and memory should slim down for newer Claude models. Practical field notes.Open Product how-tos for chat, Projects, Research, Desktop, Cowork, sharing, and account settings.Open macOS, Windows, iOS, Android, and web entry points. Same account, different reach.Open

​
Stack: skills, connectors, MCP

What MCP is, how servers expose tools and resources, and how clients connect. The socket standard underneath connectors.Open How skills are structured, installed, and triggered. Procedure packs, not access packs.Open Short product card for standing guidance that should travel with every chat.Open

​
Fluency, judgment, and safety

Delegation, Description, Discernment, Diligence. The 4D loop behind the foundations desk.Open What Anthropic expects from responsible use. Pair with your org policy before high-stakes work.Open Simon Willison’s short card on private data + untrusted content + external actions. Core agent risk model.Open Official safety notes for folder access, approvals, and what Cowork sessions can touch.Open

​
Session pocket cards

These are the one-screen reminders from class. Print or pin them. Haiku for volume, Sonnet for daily work, Opus when a miss is expensive, Fable only after Opus fails.Open Permission, human stakes, verifiability, control. Four gates before Claude gets the task.Open Built-ins, MCP, connectors, skills. Four terms, four different jobs.Open Plan, act, observe, revise. Define done, set checkpoints, read the log.Open