82 articles on workflows for builders learning Claude Code.
Claude Code in Slack lets you tag @Claude in a channel with a coding task and get back a cloud session that can open a pull request. It needs a Pro or Max plan, a GitHub repo connected at claude.ai/code, and the Claude app installed in your workspace. Team and Enterprise workspaces now use Claude Tag instead. Here is how the setup works, what happens when you tag Claude, and when I would use it over the desktop app.
Read articleClaude Code ultrareview is a deep code review that runs in the cloud. You type /code-review ultra, a fleet of reviewer agents works through your branch or pull request in a sandbox, and every finding is reproduced before it reaches you. Pro and Max accounts get three free runs, then each review bills as usage credits. Here is how to run it, what it costs, the limits, and when I think it is worth the money.
Read articleClaude Code Remote Control lets you pick up a session running on your computer from your phone or any browser. In the desktop app you type /remote-control in a session, then open it from the Claude app or claude.ai/code. Your code, files and tools stay on your machine. Here is how to turn it on, what works from the phone, how it differs from cloud sessions and Dispatch, and the limits to know first.
Read articleClaude Code can work across multiple repos in one session. Add the second project with /add-dir, keep it permanently with additionalDirectories in your settings, or add more repositories with the + button in a desktop cloud session. Here is how each option works, what an added folder does and does not load, and when to open a parent folder or move the session instead.
Read article
Claude Code security review is a built-in command, /security-review, that checks the changes on your current branch for vulnerabilities like injection, broken auth and exposed data. It reviews the diff against your remote's default branch, so it works best as the last step before you merge or ship. Here is how to run it in the desktop app, what it does and does not cover, and how to fix the origin/HEAD error.
Read article
The Claude Code goal command is /goal followed by a condition, such as all tests in test/auth pass. Claude keeps taking turns until a separate small model confirms the condition is met, judges it impossible, or an error you have to fix clears it. Here is how to set one in the desktop app, how to write a condition the evaluator can judge, how /goal differs from /loop, auto mode and a Stop hook, and what stops or pauses a goal.
Read article
Claude Code doctor is the built-in checkup you start by typing /doctor. It checks your install, finds skills, MCP servers and plugins that cost context without earning it, flags slow hooks, trims a bloated CLAUDE.md, and asks before it changes anything. Since v2.1.283 it also has /doctor prompt-audit, which reads your instructions for rules written for older models. Here is what each check does, when I run it, and how to read the report in the desktop app.
Read article
Claude Code computer use lets Claude open your apps, see your screen, click, type, and drag, the way you would, from inside the desktop app. It is off by default, needs a Pro or Max plan, and every app gets approved per session with a fixed level of control: browsers are view-only, terminals are click-only, everything else is full control. Here is how I switched it on, what it caught on a native build that nothing else could reach, which apps I refuse to approve, and the order Claude tries other tools before it touches your screen.
Read article
Anthropic released Claude Sonnet 5.5 on September 28, 2026. It costs the same as Sonnet 5, runs over 30% faster, and scored 70.6% on a test of real jobs done alone, up from 10.3%. Here are the 10 things that matter, where Opus 5.5 still wins, and how to switch today.
Read article
The Claude Code browser is the Browser pane inside the desktop app's Code tab. Claude starts your dev server, opens the app in the pane, takes screenshots, inspects the DOM, clicks through forms, and fixes what it finds, and by default it does this after every edit. It runs in a clean profile with none of your logins, which is why there is a second option, the Claude in Chrome extension, for anything that has to happen as you. Here is how the two fit together, how auto-verify behaved on a real build, and the prompts I run in the pane now.
Read article
Claude Code scheduled tasks are recurring jobs you create from the Routines page in the desktop app. Each one starts a fresh Claude Code session at the time you pick, in the folder you choose, with its own permission mode and model, and the run shows up in the sidebar for you to review. They run on your machine, so the app has to be open and the computer awake. Here is how I set up a weekday review of the previous day's commits, what went wrong in the first week, and the version of the prompt I run now.
Read article
Claude Code /clear vs /compact both free up a full context window, but they do opposite things: /clear wipes the conversation to a blank slate, /compact summarizes it so the agent keeps the thread. Use /clear between unrelated tasks and /compact when you want to keep going on the same one. Here is exactly how I decide, every session.
Read article
You can resume a Claude Code session so the agent remembers the whole thread - what you were building, the files it read, the decisions you made - instead of starting cold. Continue your most recent session or pick an older one from the list, in the desktop app or the terminal. Here is exactly how, and how to avoid the mistake that loses your context.
Read article
Claude Code project templates are pre-configured starting points that give the agent a working structure, a CLAUDE.md, and a clear scope before the first prompt fires. They cut the slow first-session setup phase and keep every project consistent. Here is the template structure I use on every build and how to install it in the desktop app in under five minutes.
Read article
Claude Code headless mode lets you run Claude as a non-interactive process - triggered by a script, a cron job, or a CI pipeline - without a human in the loop approving each step. Here is what it actually does, how to set it up safely, the difference between headless and auto-accept, and the places it earns its keep versus the places it is the wrong tool.
Read article
A technical spec for Claude Code is a plain document that tells the agent exactly what to build, in what order, and where to stop. Without one, the agent guesses at the boundaries and you spend twice as long cleaning up scope drift. Here is the spec format I write at the start of every build, the sections that actually matter, and the mistakes that make Claude Code go sideways.
Read article
Claude Code token usage is driven by how much text goes into and out of the model on every turn - your context, the files it reads, its own output, and the conversation history it drags forward. Most people burn through their limit not by doing more work, but by keeping bloated context and long sessions running. Here is exactly what uses your usage and how to stretch it.
Read article
Context engineering is the practice of deciding what Claude Code sees before it starts work - your CLAUDE.md, the files you reference, the conversation history you keep or trim. Get it wrong and Claude drifts, repeats mistakes, or writes code that contradicts your actual project. Here is the practical system that fixes it.
Read article
As of August 14, 2026, Claude Code runs in auto mode by default on Pro, Max, and Team plans instead of asking permission before every action. Here is what the safety data actually shows, what the classifier still refuses to do no matter what, and the two switches that put you back in manual control.
Read article
Refactoring is the one job where an agent can produce a beautiful diff that quietly changes what your app does. The fix is not a better prompt. It is a fence built before the first line moves, and a refactor cut into pieces small enough to check.
Read article
Claude Code reads images. Paste a screenshot of the broken layout instead of writing three paragraphs describing it. Here is how to get one in, what it can and cannot see, and the three-shot pattern that fixes things in one pass.
Read article
Claude Code is usually slow for one of three reasons: the context you are sending is too big, the tools it is running are the actual bottleneck, or you picked a heavier model than the task needs. Here is the Slow Loop Audit I run to find which one it is in about five minutes.
Read article
You can add authentication with Claude Code in an afternoon. The risk is that it looks finished and is not, because the login screen works while the server never checks anything. Here is the Auth Gate Checklist we run before any build with a login goes live.
Read article
Claude Code can write your README, your API reference, and your onboarding guide in minutes. The problem is that it will happily describe features you deleted last month. Here is the Read-Write-Prove loop I use so generated docs describe the code that actually exists.
Read article
A monorepo is the environment where Claude Code goes wrong most predictably, and the reason is not intelligence. It is scope. Here is the package-scoped session rule I use to keep an agent working on one part of a big repo without dragging the other twenty into context.
Read article
Getting a first deploy out is easy. Keeping every deploy after that from quietly overwriting someone else's work is the part nobody warns you about. Here is the deploy contract I hand Claude Code before it ships anything, and the two failure modes that cost me the most time.
Read article
How to install the Claude Code GitHub Action, the difference between interactive and automation mode, the permissions the workflow actually needs, and the cost controls that keep a runaway agent from eating your budget.
Read article
There are two different Claude code review products and people keep confusing them. Here is what the managed pull request reviewer does, what the local /code-review command does, and which one you should be using.
Read article
Claude Code reads CLAUDE.md, not AGENTS.md. If your repo standardized on AGENTS.md for another coding agent, Claude Code skips it silently - no error, no warning, just a session running with none of your project rules. Here is the one-line import that keeps both tools on the same source file.
Read article
A 400-line CLAUDE.md burns context in every session and gets followed less the longer it grows. The .claude/rules directory fixes both problems by letting you scope instructions to the files they apply to, so your API rules only load when Claude is actually reading API code.
Read article
The standalone /output-style command is gone - deprecated in Claude Code v2.1.73 and removed in v2.1.91. If a guide still tells you to type it, that guide is out of date. Here is how output styles actually work now, and how we use them for client work.
Read article
The permission prompt is not friction to switch off - it is the dial that decides how much Claude Code can do before it checks with you. Set it wrong and you either approve every keystroke or hand an agent the keys to your machine. Here is how the permission modes actually work and the setup I use for client work.
Read article
Claude Code extended thinking gives the agent extra reasoning time before it answers, and you control how much through four escalating keywords: think, think hard, think harder, and ultrathink. Here is what each level actually buys you, and the CCC Effort Ladder I use to decide when it is worth the wait.
Read article
Claude Code runs on a usage allowance that resets on a rolling window, and the worst time to learn that is mid-build. Here is what actually happens when you hit the limit and the habits I run so it stops catching me off guard.
Read article
A statusline is the one line at the bottom of Claude Code that can tell you the model, the branch, the context left, and the cost - without asking. Here is how I set mine up so I stop losing track of long sessions.
Read article
Claude Code checkpoints every prompt automatically and /rewind can undo a bad session in seconds. Here is exactly what it does, the specific things it can't undo, and why I still commit to Git on every real build.
Read article
Claude Code's built-in sandbox restricts what Bash commands can read, write, and reach on the network, enforced by the operating system. It's the single best safety upgrade for running Claude Code with less hand-holding, but it protects less than most builders assume. Here's exactly what it covers.
Read article
Claude Code context management explained: how the context window works, what breaks it, and the exact habits I use to keep large projects running cleanly without the agent losing the plot.
Read article
Writing a project brief for Claude Code before you open the app cuts build time in half. Here is the exact brief format I use for every client project and internal build.
Read article
Git worktrees let you run two Claude Code sessions on the same repository at the same time, each in its own directory and on its own branch. No file conflicts, no blocked agents waiting, no merge headaches mid-build. Here is how to set them up in two minutes.
Read article
Building a dashboard with Claude Code starts with the data, not the design. Tell Claude Code what numbers you need to track, paste in a sample of your data, and it generates the component structure, wires in the charts, and handles the responsive layout. You direct; it builds.
Read article
Most Claude Code beginners jump straight to building. The ones who get better results start with a planning pass first. Here is the plan-first workflow I use on every project above a few hours, and how to prompt Claude to think through the whole build before it writes a single line of code.
Read article
Throwing an error at Claude and hoping it fixes it is not a debugging workflow. Here is the three-step method that actually works: describe what you expected, paste the full error, and tell Claude what you already tried. Thirty seconds of setup cuts debugging time in half.
Read article
Claude Code writes code fast, and without a commit rhythm you can lose hours of working progress in one wrong command. Here is the branch strategy, commit pattern, and four rules I follow on every build to keep AI-written code clean and recoverable.
Read article
Claude Code does not remember your project between sessions by default. Every new conversation starts from scratch. The memory bank changes that - here is how to use CLAUDE.md and a few supporting files to give Claude permanent context on any project.
Read article
The right way to manage .env files with Claude Code is to keep secrets out of the conversation entirely. Set up your environment variables before the session starts, reference them by name in prompts, and never paste a live key into the chat. Here is the Safe Env Pattern I use on every project.
Read article
Running parallel Claude Code agents means deliberately opening two or more sessions to work on separate tasks simultaneously - not waiting for one to finish before starting the next. The Fork Method is the system I use to decide when to split work across sessions and how to keep them from colliding.
Read articleClaude Code forgets everything when a session ends. The Memory Bank is a two-file system that gives it persistent context, so every new session starts informed instead of blank. Here is how to set one up in ten minutes.
Read article
Leaving Claude Code to work on its own overnight sounds like the dream. I tried it on an actual client build and learned exactly where it saves you hours and where it quietly wanders off. This is the Leash Loop I use to get unattended work without waking up to a mess.
Read article
Skills and MCP servers both extend Claude Code, and people mix them up constantly. They solve different problems. Here's the plain-English difference, the Skill-or-Server Test I run before building either, and the mistakes that make people reach for the wrong one.
Read article
Most people open Claude Code, poke around, and never get it configured to actually help them. This is the Day-One Stack I walk every new member through - the plan to buy, the desktop app choice, and the two files that make Claude useful from your first real session instead of your tenth.
Read article
The fastest way to stop Claude Code from confidently shipping broken code is to make it write the test first. When the test exists before the code, the agent has a target it cannot fake its way past. Here is the Red-Green-Verify loop I use to get working code on the first pass.
Read article
Subagents let Claude Code spin up separate agents that each own a task and run at the same time. Used well they cut a multi-part build from an afternoon to an hour. Used badly they make a mess in parallel. Here is the Delegate-Verify loop I use to get the speed without the chaos.
Read article
CLAUDE.md is the project memory file Claude Code reads at the start of every session. Set it up once with your conventions, guardrails, and always/never rules and Claude walks in already briefed - no re-explaining, no drift, no wasted back-and-forth on every new chat.
Read article
Slash commands are saved prompts you trigger by typing a slash and a name - a way to run your best multi-step workflow in one keystroke instead of retyping it every time. Here are the custom commands I built for client work, exactly how I wrote them, and the ones that actually earn their place in my day.
Read articlePlan Mode is Claude Code's read-only research mode - it reads your codebase and writes a step-by-step plan for your approval before it touches a single file. Here's the exact loop I run on every client build to catch the wrong approach in a plan I can edit in seconds, instead of in code I'd have to rip out an hour later.
Read article
Most freelance deals die not because of a 'no' but because of silence. This guide covers the exact cadence, messages, and mindset you need to follow up with a client without feeling pushy - and without losing the deal to someone who stayed in touch.
Read article
Claude Fable 5 ran a two-month codebase migration in a single day in testing. That does not mean you point it at your legacy system and hope. Here is the four-phase method for migrating real code with Fable, keeping behavior intact and your client's trust with it.
Read article
Claude Fable 5 costs roughly double the Opus class per token, and in long agent runs those tokens compound fast. The fix is not avoiding Fable, it is using it deliberately. Here are five levers that cut the bill hard while keeping the results that made you reach for it.
Read article
Claude Fable 5 is built for long-horizon agents, the kind that plan, act, and keep going across millions of tokens. But the model is the easy part. Here is the contract every Fable 5 agent needs to actually finish the job instead of looping forever or quietly making things up.
Read article
Claude Fable 5 vs Opus 4.8 is not a close fight on paper, but it is the wrong question for most builds. Here is the honest split: what Fable's Mythos-class tier actually buys you, where Opus 4.8 is still the smarter call, and a simple two-question test for picking the right one every time.
Read article
A hands-on Claude Fable 5 review after putting the first public Mythos-class model through real work. What genuinely impressed me, where it frustrated me, and a straight answer on whether it is worth switching to right now, or whether you should wait.
Read article
Anthropic shipped Claude Fable 5, pulled it three days later under a US government export-control order, and redeployed it globally on July 1. Here is what actually changed, why it vanished and came back, and how to put the most capable public Claude model to work on a real build right now.
Read article
A Claude Code hook is a shell command the app runs automatically at set moments in a session, like right after an edit or before a tool runs. Here is what hooks are, the four moments they can fire, and two you can copy today to auto-format your code and protect files you never want touched.
Read article
A subagent is a second Claude Code instance that runs a job in its own context window and reports back. Here is what Claude Code subagents are, the three jobs they do well, how to spawn and build your own, and what the June 2026 update changed.
Read article
Juggling several clients in Claude Code does not have to mean a dozen scattered windows. Here is the client cockpit method: one repo and one CLAUDE.md per client, one view to see every job, and the guardrails that keep client work from bleeding together.
Read article
Anthropic launched Claude Tag, a new way to put Claude inside Slack as a standing teammate you tag like any coworker. Here is what it does, who can use it today, and the standing-agent shift it points every builder toward.
Read article
The best Claude Code MCP servers are the ones that connect the agent to the tools and data you already use. Here are the ones worth installing, what each unlocks, and how to add them without bloating your setup.
Read article
A practical starter set of skills that earn their place on real work. Each one is described by what it does and when it pays off, so you install the ones that match your actual workload, not a pile you never use.
Read article
A practical guide to scheduling recurring Claude Code agents - a morning report, a nightly cleanup, a weekly research pull - and the one rule that keeps unattended runs from costing you a client.
Read article
Auto mode lets Claude Code take a whole sequence of actions while you watch from checkpoints. Here is what it actually changes, and how to use it without burning money or breaking something you cannot undo.
Read article
Four primitives, one decision. This is the plain-English guide to Claude Code skills, subagents, MCP, and hooks, with a table that tells you which to reach for and the mistakes that quietly slow your work down.
Read article
Most people stall at Level 1 and never realize there are four more above them. Here is the full progression, how to self-diagnose where you are, and the one concrete move that gets you to the next level.
Read article
Claude Code can now write its own harness for a task, fan the work across subagents, and run it in the background while your session stays free. Here is how to turn dynamic workflows on and run your first one today.
Read article
Your Claude Code sessions keep forgetting what you're building because you only set up one layer of memory. There are three. Here's the stack the shippers are running this month - and how to install it in a weekend.
Read article
Every business running Claude Code is bleeding context. They don't know it's a problem yet. The shops that show up with a productized memory-engineering deliverable are about to spend twelve months printing $5K retainers. Here's how I'm pricing it.
Read article
The headlines led with the valuation. The part that actually matters: the model sitting under your Claude Code session is about to get longer, cheaper, and faster shipped. Here's what the milestone unlocks for builders, and the habit stack to install this week so you ride it.
Read article
The client that pays the most isn't always the client that earns the most. I cut my biggest invoice in May, and the agency's margin doubled by the end of June. Here's the math that made the call obvious - and the playbook for spotting your version of the same client.
Read article
Most builders learn Claude Code. The ones who ship have a different set of habits. Here are seven of them - the ones that take you from tinkering forever to actually shipping the thing this week.
Read article
The fastest way to cap your agency's earnings is to charge by the hour for something Claude Code builds in a weekend. Here's the pricing system top AI shops actually use - and why the unit you sell matters more than the rate.
Read article
Inside the club, a daily rhythm has emerged that's quietly giving builders back half a working week. Here's the workflow - morning planning, slice-based builds, agentic test loops, end-of-day commits - broken down by where the hours come from.
Read article
Learning Claude Code is fun, but turning it into income is the goal. Here are five concrete, repeatable workflows our members use to land clients and ship paid work.
Read articleWe use cookies to keep the site running, remember your preferences, and understand how people find us. See our Cookie Policy for details.
David Iya
Duncan Rogoff