Getting started with Claude Code: from install to your first request
Install the coding agent that runs in your terminal and hand it your first task.
Assumes basic terminal skills
EXPLORE ARTICLES
Articles & dev notes
Implementation guides, design thinking and troubleshooting. Find them by context and code.
ARTICLES & NOTES
22 ARTICLESArticles & dev notes
Writing CLAUDE.md: stop explaining things every time
Where to put it, how to write it, how long it should be, and what to use when you need to enforce a rule.
For people who have used Claude Code once
Permissions and modes: deciding what to hand over to Claude Code
Permission modes that set how often it asks, the evaluation order of allow/ask/deny rules, and the wildcard pitfall.
For people who have used Claude Code once
Practical workflows: investigate, fix, and split the work
Plan mode, subagents, worktrees and PR creation: a way of working that follows everyday tasks.
For people who have used Claude Code once
Deep learning basics: how a model learns a function from data
Neurons, layers, loss, gradients and backpropagation. Teach a small network XOR and see, briefly, how deep learning learns.
Enough if you can read Python
How LLMs work: predicting the next token
Tokens, probabilities, temperature and the context window. Run a toy model to see how an LLM builds an answer, and why it gets things wrong.
Enough if you can read JavaScript
A look inside the Transformer: computing attention by hand
The Transformer is what computes next-token probabilities inside an LLM. See its overall structure, and compute its core, attention, in a small piece of code.
For readers of “How LLMs work”
LLM models at a glance: kinds and how to choose
Cloud APIs and open models, size, MoE, thinking, multimodal and roles. A brief map of how to sort the many LLM models and how to pick one.
Enough if you have used an LLM
I tried to make Claude Code read my .env: a Read deny rule still leaked through grep -r and Python
An experiment making Claude Code actually read a .env with fake keys. CLAUDE.md was obeyed and a Read deny stopped cat, but grep -r and Python slipped past it. Only the sandbox stopped everything.
For people who have used Claude Code once
The length of "👨👩👧" is 8 in JavaScript, 5 in Python, 18 in Rust and 1 in Swift
The same string has a different “length” in each language. I counted 𠮷野家, a flag and a family emoji in four languages, to see why, and how to handle them without breaking anything.
If you can read any one of the languages, that’s enough
Local LLMs: running a model on your own machine
What a local LLM is and how it differs from a cloud API. Estimate memory, understand quantization, run a model with Ollama, and know where it fits.
Enough if you can use a terminal
Running local LLMs seriously: big models and the KV cache
How to size the memory beyond the weights (the KV cache), split a model across GPUs, handle MoE, serve several users, and choose a serving tool.
For readers of the local LLM introduction
Build a system that answers questions about your documents (local LLM × RAG)
Split documents, search for the relevant parts, and hand them to a local LLM to answer. Build a question-answering system for internal documents in dependency-free Python.
Enough if you can read Python
Running internal-document Q&A in production: permissions, updates, evaluation and operations
Turn document Q&A (RAG) into a production system many people use. Incremental ingestion, permission filtering, better retrieval, evaluation and operations.
For readers of the basic RAG article
Getting started with AWS: the big picture and a safe start
Regions and AZs, the main services, a typical web architecture and the shared responsibility model. Plus the safety settings to make first, and the cost pitfalls.
Enough if you know basic IT
A stack for solo developers and early startups: build small, keep operations light
How to choose technology to build and run with a small team, using the setup of a real web app (React Router, Cloudflare Workers, D1, Clerk) as the example.
Enough if you have built one web app
Trying LLMs on Colab: run a model on a free GPU
Even without a GPU in your own PC, Google Colab's free GPU can run an LLM. Two ways to run one (Ollama and Transformers), what you may use it for, what is prohibited, and how to size the memory.
Enough if you can run Python code in a notebook
Mental health for engineers: what you can do before you burn out
How to recognize burnout, a way to notice early and step back in small ways, and where to turn so you don’t carry it alone.
For people who are running out of slack
git worktree: one working folder per branch
How to open several branches at once from a single repository, without stashing or cloning.
For people comfortable with basic Git commands
Using your own PostgreSQL from Cloudflare Workers: connect without opening a port, and share one server between services
Full-text and vector search led us to run PostgreSQL on a VM behind a Workers site. A record of connecting it through Hyperdrive, Workers VPC and Tunnel, separating services with roles, and keeping backups and batch jobs on the VM.
For people who have touched Workers and a little PostgreSQL
IN (subquery) over UNION ALL read 570,000 rows: PostgreSQL's estimates, and = ANY(ARRAY(...))
After two tables were shown as one through UNION ALL, a product-code search took 4 seconds. Reproduced locally, IN made the planner walk the whole popularity index. A record of reading the estimates to find the cause, and getting back to 0.3 ms with = ANY(ARRAY(...)).
For people who have looked at EXPLAIN
Keeping an AI agent out of production: the AI writes code; a person merges, runs production and types secrets
A record of taking a change from database to production deployment with Claude Code. The AI wrote code and checked it locally; a person merged, ran production and entered secrets. What actually went wrong, and why it still did no harm.
For people starting to use Claude Code on a real service