How do I make Claude run without me sitting there?

Three mechanisms, each with a different trigger. A hook fires on an event — after every edit, before every command. A schedule fires on the clock — every Monday at 9am. A loop repeats on an interval until the condition you set is met.

Does Claude use my logged-in accounts?

Yes — it shares your browser's login state, so it can work inside any site you're already signed into without separate API access. That's what makes it useful for internal portals and CRMs, and also why you should be deliberate about which sites you grant it.

Can I see what it's doing?

Yes. Browser actions run in a visible window in real time, so you watch the tabs open and the fields fill. When Claude hits a login page or a CAPTCHA it stops and asks you to handle that part yourself.

Google Chrome and Microsoft Edge are supported, and Claude Code also connects to other Chromium-based browsers including Brave, Arc, Vivaldi, and Opera. It isn't supported inside Windows Subsystem for Linux (WSL).

Do I need a paid plan?

Yes. Browser integration requires a direct Anthropic plan — Pro, Max, Team, or Enterprise — and a session signed in with your Claude account. It isn't available through third-party providers like Amazon Bedrock, Google Cloud's Agent Platform, or Microsoft Foundry.

Does /loop keep running after I close my terminal?

No. /loop runs inside your current session — when the session ends, the loop stops. For work that needs to run without you, use a scheduled agent.

Can a scheduled agent do everything a /loop does?

In principle, yes — but a scheduled agent is heavier to set up. For 'check this every 5 minutes while I'm working,' /loop is much simpler. For 'do this every Monday,' a scheduled agent is the right tool.

How do I see what's scheduled?

Use /schedule to list scheduled agents. Active loops are visible in your current session.

Can a hook do what a skill does?

No. A hook runs a shell command — it can't change how Claude thinks or writes. A skill shapes Claude's behaviour and knowledge. They're complementary, not interchangeable.

Should I use a hook or a skill to enforce code style?

Both, for different aspects. A skill tells Claude 'write code in this style' (shaping what it produces). A hook runs your formatter after every edit (enforcing the mechanical rules). The skill prevents style issues; the hook catches anything that slips through.

Can hooks block Claude from doing something?

Yes. A PreToolUse hook can reject an action before it happens — like blocking any edit to the main branch. That's guardrail territory, not knowledge territory, which is exactly what hooks are for.

Hooks, schedules, and loops — Claude for Humans

Everything on this page answers the same question — what makes Claude start working when you are not the one asking? — and the only real difference between them is the trigger. Sort them by that and the whole area stops being a pile of similar-sounding features.

An event trigger is a hook: something happens in the tool, and your command runs. A clock trigger is a scheduled agent: it runs at 8am whether or not you are awake, and it outlives the session you set it up in. An interval trigger is a loop: the same job re-runs every few minutes while you work, and stops when you close the session. And a stream trigger is Monitor, the odd one out — Claude watches a live output (a test runner, a deploy, a log tail) and reacts the moment a line matters. /goal is not a trigger at all; it is the finish line you attach to any of them so Claude keeps going instead of stopping at its first reply.

The two comparisons below exist because these are the two pairs people actually mix up: schedule versus loop (a calendar appointment versus an in-session timer), and hook versus skill (a command that fires versus knowledge that shapes how Claude works). Neither pair is interchangeable, and picking the wrong one usually shows up as either over-engineering or an automation that quietly stopped running.

Worth being honest about the surface here. Scheduling on Desktop is a plain-language ask — “every Monday at 9am, draft the team update” — with no cron syntax to learn. Hooks are not: a hook runs a shell command and lives in a settings file, which makes it a Power-Track and team-setup tool rather than something you reach for in the everyday chat.

The one rule that matters more than any of the mechanics: automate only what you already trust Claude to do while you are watching. Unattended work multiplies whatever you set up, including the mistakes, and nobody is there to catch the first one. This is where permissions stop being a formality — the boundaries you set are the only supervision the job gets.

How it works in Claude

Hooks Small commands that fire automatically at set moments — like running your formatter every time Claude saves a file. Scheduled agents & loops Hand Claude a job and a time, and it runs itself — every morning, every Monday, or on a repeating loop — without you in the room. Monitor (watch & react) Claude watches a running command's output live and reacts the moment something happens — an error, a passing test, a new log line.

Commonly confused with

Two things people mix up here, and the difference that actually matters.

Scheduled agent vs /loop in Claude Code Both repeat work automatically — but a scheduled agent is fire-and-forget on a calendar, while /loop is an in-session timer. Pick the wrong one and you're either over-engineering or under-automating. Hook vs Skill in Claude Code Hooks and skills both happen automatically — but a hook runs a command at a fixed moment, while a skill shapes how Claude thinks. Mixing them up leads to the wrong tool for the job.

Claude in Chrome: letting Claude use your browser Most of the tedium in knowledge work isn't in your files. It's in a browser tab — the CRM, the portal, the form you fill in forty times. This is the feature that reaches in there. 10 min

How to do it

/hooks Set up commands that run automatically at certain moments.
/schedule Schedule a cloud agent to run once or on a recurring cron — even when you're not around.
/loop Run a prompt or slash command on a repeating interval until you stop it.
/goal Set a lasting goal and let Claude keep working and re-checking until it's met.

How it works in Claude

Commonly confused with

Read more

How to do it