How to organise your AI workspace repo vs individual client project repos. What lives where and why.

How the GitHub MCP server connects Claude Code to your repos. Auth flow, scopes, and what you unlock once connected.

Three-location model, global vs project scope, the 7-section skill template, and how to write a skill from scratch.

How this chat and Claude Code work together. Plan and generate here, build and execute in the terminal.

Every /ce: command explained — what it does, when to use it, and how it relates to the others. Organised by feature lifecycle phase with a quick-reference decision table.

A no-IDE setup using only plan.md files and voice. Four-layer flow: input → brain → shared glue → output. How to run Claude Code without ever opening an editor.

Replace chaotic terminal tabs with three layers: tmux for named windows, git worktrees for isolation, and Agent Teams for coordinated parallel agents. Visual diagrams and daily workflow.

How /wave-execute fights context rot by splitting plans into dependency waves and dispatching each task to a fresh 200K-context agent. The full pipeline from ce:brainstorm to deployed code.

Where every skill fires, which hooks run underneath, and how /resume + /ship close the loop between sessions. Covers discovery, Adoption Discipline triage, planning, TDD build, verification, the verified ship procedure, and session-boundary continuity.

Claude writes memory files locally during sessions. A Stop hook auto-commits them to GitHub so every new session starts with full context — across machines and accounts.

Architecture decisions, system flows, mental models — anything worth visualising can be added here.