Short-form references, concepts, and working notes.
16 notes, newest first.
The discipline of designing everything around an AI model that makes it reliable in production
The discipline of filling an LLM's context window with the right information for each step, replacing prompt engineering as the core AI skill
Multi-session orchestration where independent Claude instances coordinate through shared tasks and direct messaging
Modular knowledge packages Claude loads on-demand when your task matches their description
Shell commands that run at specific points in Claude Code's lifecycle for automation and validation
Complete reference for built-in and custom slash commands in Claude Code
Autonomous AI coding technique using persistent iteration loops until task completion
Isolated AI workers in Claude Code with their own context windows, tools, and permissions
Development methodology where specifications become the primary artifact and code is generated from them
Research framework for crafting and evaluating constitutional principles that actually work in AI alignment
Anthropic's training methodology that uses AI feedback instead of human feedback to align models
Extracting implicit principles from preference data by inverting the Constitutional AI process
Open standard for connecting AI models to external tools, data, and systems via a unified interface
Democratic approach to AI alignment where populations source principles through deliberation
The Heath brothers' SUCCESs framework for making ideas stick
“The first principle is that you must not fool yourself, and you are the easiest person to fool.” “Scientific knowledge is a body of statements of varying degrees of certainty: some most unsure, some nearly sure, but none absolutely certain.” “Permit us to question, to doubt, to not be sure.” “Have...