Learning center
Learning to build with AI coding tools
Working out how these tools behave, and how to get more out of them. These pages explain what AI coding is good at, where it stops working, and the habits that make the difference. They are ordered by where the work happens, from your first session to running several agents on a machine of their own.
Start here Guide · 10 min
Set up Claude Code on your computer
Never opened a terminal? This guide takes you from nothing to Claude Code running on your own computer, and lets Claude install Git, GitHub and Docker for you.
Start the guideWhat these tools are
What AI coding actually is, what it is good at, and the habits that decide whether you get a good result.
- Article 5 min read What vibe coding is What vibe coding is, how it works in practice, why the first version comes out so well, and the four places generated code quietly does the wrong thing.
- Article 5 min read Getting better at AI coding Vibe coding best practices that hold up: describe the contract, work in checkable steps, write context down, review at the seams, keep feedback fast.
Workflows
Set up a session, review its changes, and coordinate independent tasks across several agents.
- Article 5 min read Claude Code terminal workflow Run Claude in the terminal as a workflow, not a chat window: a branch per task, a project instruction file, sandboxed permissions and reading the diff.
- Article 5 min read Running multiple agents How to run multiple agents in Claude Code: a git worktree per session, tmux to keep them alive, issues as the shared queue, and when one agent wins.
Connecting it to your setup
Giving the agent access to the tools, data and services your team already uses, without paying for connections you do not need.
- Article 5 min read What MCP servers do What MCP servers do in Claude Code: the tools and data they hand the agent, local, project and user configuration, and the context and trust they cost.
- Article 5 min read Connecting Claude Code to your tools Connecting Claude Code to your tools: when a command line tool such as git or gh is all you need, and when a job really calls for an MCP server.
Where the agent runs
Use containers for a repeatable environment, or a remote machine for sessions that need to keep running.
- Article 6 min read Claude Code in a container What running Claude Code in a container actually changes: what the agent can reach, why teams do it, where secrets still matter, and what isolation costs.
- Article 5 min read Claude Code on a remote machine How to run Claude Code remotely on a server, home lab box or VM: persistent tmux sessions, reconnecting from elsewhere, credentials and Windows.
Ready to put something into production?
The resources section covers hosting, databases, access and recovery for applications people actually depend on.