Constellation · The Navigator

Claude Code: Master the agentic coding loop

Twenty-eight guides from first session to measured harness improvement, covering project memory, context, skills, hooks, permissions, verification, and longer coding workflows.

28 of 28 guides live · fact-checked against official docs

Chapter 0

☆

Start Here

Install Claude Code, run your first agentic session, and understand the loop behind it.

9 minbasics / install / mental model

Chapter 1

☆

CLAUDE.md

Write a CLAUDE.md that makes every session smarter — and keep it from rotting.

11 minmemory / configuration / best practices

Chapter 2

☆

Context

Control what fills the context window and keep long sessions sharp.

10 mincontext / performance / sessions

Chapter 3

☆

Skills

Package workflows as skills Claude discovers and applies on its own.

11 minskills / automation / customization

Chapter 4

☆

Commands

Build custom commands and tune the interface until it disappears.

8 mincommands / customization / workflow

Chapter 5

☆

Subagents

Fan work out to specialist agents and keep your main thread clean.

12 minagents / delegation / architecture

Chapter 6

☆

Orchestration

Coordinate fleets of agents — workflows, teams, and manager patterns.

12 minorchestration / teams / scale

Chapter 7

☆

Hooks

Enforce rules with code, not hope — gates that fire every time.

12 minhooks / automation / guardrails

Chapter 8

☆

MCP

Wire Claude into your tools, data, and services through MCP servers.

11 minmcp / integrations / tools

Chapter 9

☆

Permissions

Run Claude with exactly the autonomy you intend — no more, no less.

10 minsecurity / permissions / configuration

Chapter 10

☆

Workflows

Ship multi-step changes with plans, tests, and tight feedback loops.

12 minplanning / tdd / workflow

Chapter 11

☆

Autonomy

Let Claude work unattended — safely, with the right harness.

11 minautonomy / automation / headless

Chapter 12

☆

Models

Pick the right model for the task — capability, speed, and cost.

9 minmodels / selection / pricing

Chapter 13

☆

Cost & Speed

Get more done per token and per dollar — without losing quality.

10 mincost / performance / optimization

Chapter 14

☆

Rescue

Diagnose the common failure modes fast — and know where answers live.

8 mintroubleshooting / faq / help

Chapter 15

☆

Plugins

Bundle skills, agents, hooks, and MCP into one installable plugin — and ship it to your team.

11 minplugins / distribution / customization

Chapter 16

☆

Agent SDK

Drive the same agent loop from Python or TypeScript — typed messages, programmatic permissions, hosting.

12 minsdk / automation / advanced

Chapter 17

☆

Checkpoints

Undo a bad turn, branch a session, and recover work without leaning on git.

9 minsessions / recovery / workflow

Chapter 18

☆

Surfaces

Move one session across terminal, IDE, desktop, browser, and phone — and pick the right surface.

10 minsurfaces / ide / remote

Chapter 19

☆

All Commands

Every built-in slash command, keystroke, and prefix in one grouped map — so you stop guessing what exists.

9 mincommands / reference / cheatsheet

Chapter 20

☆

Experience

The UI, voice, and personality commands nobody documents — what each does and when it actually helps.

7 mininterface / voice / customization

Chapter 21

☆

GitHub & CI

Stop conflating the four ways Claude reaches GitHub, and wire up the right one.

11 mingithub / ci / github-actions / code-review / automation / claude-code

Chapter 22

☆

Teams & Enterprise

Deploy and govern Claude Code across an org — the policy stack, Bedrock/Vertex, analytics, and security controls.

11 minclaude-code / enterprise / teams / governance / managed-settings / bedrock / vertex-ai / security

Chapter 23

☆

TDD & Debug

Give Claude a check it can run, then let the loop close itself to green.

11 mintdd / debugging / verification

Chapter 24

☆

Prompting

Write prompts that close the loop on the first pass — outcome, context, and a check Claude can run.

13 minprompting / best practices / workflow

Chapter 25

☆

Loop Engineering

Compose /goal, /loop, hooks, subagents, and state files into production-grade agent loops.

21 minloop engineering / goal / automation / hooks / subagents

Chapter 26

☆

Harness Config

Build a repo-specific Claude Code harness with instructions, permissions, hooks, skills, subagents, MCP, and verification.

18 minharness engineering / configuration / claude-code / permissions / skills

Chapter 27

☆

Recursive Self-Improvement Guide

Build a candidate skill, replay matched software tasks, and judge it against product behavior.

21 minclaude-code / recursive improvement / skills / evals