Guyaga
WorkServicesAssetsWritingAbout
Start a project

Guy Aga

HomeWorkServicesAssetsWritingAbout
  • YouTube
  • LinkedIn
  • Instagram
  • Facebook

All writing · 17 Mar 2026 · 5 min read

The Claude Code Commands I Use Every Day (And You Should Too)

Anthropic shipped a wave of new slash commands for Claude Code — /effort, /loop, /copy, /color, /simplify, /batch, and /reload-plugins. Here's how each one works and fits into a real daily workflow.

In this article

  1. /effort — Control How Hard Claude Thinks
  2. /loop — Automate Recurring Prompts
  3. /copy — Pick What You Want to Copy
  4. /color — Color-Code Your Sessions
  5. /simplify — Parallel Code Review Agents
  6. /batch — Parallel Migrations at Scale
  7. /reload-plugins — Hot-Reload Without Restarting
  8. My Daily Setup

Anthropic is shipping Claude Code updates so fast it's hard to keep up. Over the past few weeks, we've gotten a wave of new slash commands that change how you manage effort, automate tasks, copy code, and even color-code your sessions.

Here's my breakdown of every new command, when to use each one, and how they fit into a real daily workflow.

New feedback prompt in Claude Code


/effort — Control How Hard Claude Thinks

This is the one I use most. Before /effort, adjusting Claude's reasoning depth was buried inside the /model menu. Now it's a direct command.

/effort slash command overview

Type /effort and pick your level:

Setting effort level

  • Low — Skips deep reasoning. Fast and cheap. Great for renaming variables, formatting, simple edits.
  • Medium — Balanced. The default sweet spot for daily coding work.
  • High — Full reasoning power. Use for architecture decisions, complex debugging, multi-file refactors.

To reset: /effort auto — lets Claude decide based on task complexity.

/effort auto mode

The key improvement: /effort now works while Claude is responding. You can switch effort levels mid-conversation without waiting for the current response to finish.


/loop — Automate Recurring Prompts

/loop lets you run any prompt or slash command on a recurring interval. Set the time, give it a task, Claude repeats it.

/loop command

code
/loop 5m check the deploy

/loop 5m example

That checks deploy status every 5 minutes. You can pair it with any command:

code
/loop 10m /stats

/loop 10m /stats

This runs /stats every 10 minutes to track token usage during long sessions.

Best use cases:

  • Monitoring a running deployment
  • Checking test results after changes
  • Watching for build failures in CI
  • Running periodic health checks

Pro tip: /loop works best for lightweight, read-only checks. Running heavy code generation on a loop will burn through tokens fast. Use it for monitoring — not for repeated builds or refactors.

To kill all loops: set CLAUDE_CODE_DISABLE_CRON=1.


/copy — Pick What You Want to Copy

Before /copy, getting code out of Claude responses meant scrolling and manual selection. /copy adds an interactive picker.

/copy command

When Claude's response has code blocks, /copy lists each one individually:

/copy code block picker

Select a specific block or grab the full response. Options include:

  • Individual code blocks (component, API route, test file, etc.)
  • Full response

SSH bonus: Press w on the focused selection to write directly to a file — bypasses clipboard entirely. If you've ever fought clipboard issues in tmux or remote sessions, this is a lifesaver.

"Always copy full response" mode: Select it from the picker, and future /copy commands skip the block menu entirely. Reset anytime through the picker.


/color — Color-Code Your Sessions

Simple but powerful when running multiple terminals.

/color command

/color red sets a red prompt bar for the current session.

/color red in action

Running three terminals — frontend, backend, infrastructure? Give each a different color. At a glance, you know which session you're in.

Reset with any of: /color default, /color gray, /color reset, /color none.

/color default reset

Pairing with /rename

/color works great alongside /rename. The session name now shows on the prompt bar — combine a color with a descriptive name like "auth-service" or "frontend-v2" for effortless session management.

/color + /rename combo


/simplify — Parallel Code Review Agents

This is one of the most underrated commands in Claude Code. /simplify is a post-implementation cleanup tool. Build your feature, get it working, then run:

code
/simplify

/simplify command

It checks git diff to find what changed, then launches three specialized agents in parallel:

  1. Code reuse agent — Finds duplicate logic and consolidates it
  2. Code quality agent — Catches code smells and enforces your CLAUDE.md rules
  3. Efficiency agent — Flags performance issues and structural improvements

All three run simultaneously. When done, /simplify aggregates findings and applies fixes in one pass.

You can focus it: /simplify focus on error handling


/batch — Parallel Migrations at Scale

If /simplify is your code quality pass, /batch is your migration engine. It handles the kind of work that usually means a long afternoon of find-and-replace across dozens of files.

/batch command

code
/batch migrate src/ from Jest to Vitest
/batch replace all uses of lodash with native equivalents
/batch add type annotations to all untyped function parameters

How it works:

  1. Research & Plan — Explores the entire directory, scans every file, produces a migration table. Presents the plan and waits for approval before touching any code.
  2. Execute in Parallel — Decomposes work into independent units, spawns agents in isolated git worktrees.
  3. Verify & PR — Each worker runs verification checks and /simplify before creating PRs. Every pull request comes out pre-reviewed.

Key difference: /simplify = quality check on recent changes. /batch = large-scale migrations across many files. /batch thinks first, shows the plan, and waits for your approval.


/reload-plugins — Hot-Reload Without Restarting

If you build or use Claude Code plugins, this removes a major friction point.

/reload-plugins command

Previously, updating a plugin meant restarting your session — losing context, re-establishing state, waiting for reload. /reload-plugins activates pending changes immediately. Edit your plugin, run the command, changes take effect.


My Daily Setup

After using all of these for a few weeks, here's how they fit into my actual workflow:

  • /effort — I toggle this constantly. Low for quick fixes, high for architecture work.
  • /copy — Multiple times per session. Grabbing specific code blocks is so much faster now.
  • /color + /rename — Every session gets a color and name. Makes multi-terminal work manageable.
  • /simplify — After every feature. It's become a habit like running tests.
  • /loop — Occasionally, but powerful when monitoring deploys.
  • /reload-plugins — Essential when developing custom skills.

These commands didn't arrive as one big release — they were spread across multiple versions over a few weeks. That's exactly why most people missed them.

AI · Claude Code · Tutorial · Productivity

Read next

The Nano Banana Pro Prompting Guide: 6 Techniques With Real Examples

Newer

Claude Skills 2.0 — From Recipe to Kitchen

Let’s build something.

Only I read it. Privacy

Or write to me

Guy@aga.digital
  • YouTube
  • LinkedIn
  • Instagram
  • Facebook

Guy Aga · Tel AvivPrivacy