Skip to content
← All guides

The 3 Best Claude Mods (+10 More)

Here are the 3 prompts from the video, plus 10 more. Paste a prompt into Claude, and it builds the mod for you. No coding.

Before you paste anything

A mod is a small add-on that changes how Claude looks and works. You describe it in plain English and Claude builds it for you.

  • Where they run The Code tab in the Claude desktop app
  • Regular Claude chats Mods don't show up there
  • Coding needed None
  1. Open the Code tab. Open the Claude desktop app, click Code, and pick any folder.
  2. Paste a prompt and send it. Claude writes the mod for you.
  3. Click "Turn them on." Claude asks once if you want to turn on the mods it writes. If you say no, the mod won't show up until the next time that session starts.
  4. Type the code. Every mod here gets its own code, like /privacy. It only works after Claude has built the mod, so paste the prompt first.
  5. Ask Claude how to try it. Every build comes out a little different, so have it tell you exactly what to type or click.

Don't like something? Just say so. "Make the bar smaller" or "stop flagging my business name" works, and Claude updates the mod for you.

The 3 from the video

1. Remember This: Claude never forgets what you teach it

Every time you correct Claude or tell it something important, this catches it and decides where it belongs. A rule for how Claude works goes into your CLAUDE.md, so every future chat follows it. Something to remember about you or your work goes into your second brain. You press Save, and you never explain it twice. Your code is /remember-this.

Prompt 1: Remember This
Make me a Claude mod that makes sure Claude never forgets anything I teach it.

Every time I correct Claude, tell it how I like something done, or share something important about me, my work or my business, catch it. Then decide where it belongs:

1. A rule for how Claude should work (like "never use em dashes" or "always ask before deleting files"). Save it as a permanent rule in my CLAUDE.md file so every future chat follows it.
2. Something important to remember about me or my work (like my business name, a client's preferences, or a decision I made). Save it as a note in my second brain.
3. Both, if it's both.

Show what you caught in a small bar above the prompt, with where you want to save it and why, and buttons for Save, Edit and Skip. Never save anything until I press Save. Don't save the same thing twice, and if a new rule contradicts an old one, show me both and let me pick.

The first time it runs, ask me where my second brain lives (an Obsidian vault, a notes folder, or nowhere yet). If I don't have one, make a folder called second-brain in my home folder and save notes there as simple markdown files.

Name it /remember-this, so I can also type that after anything I want saved right now, and /remembered to see everything it has saved.

Idea from reflect-mod by BayramAnnakov, upgraded to save into your second brain too.

2. Command Explainer: know what you're approving

Claude asks for permission all day, and most of it reads like gibberish. This adds one plain-English line to every permission prompt: what it will do, what it touches, and a green, yellow or red risk level. Your code is /explain-commands.

Prompt 2: Command Explainer
Make me a Claude mod that explains every command in plain English before I approve it.

Whenever Claude asks for my permission to run a command, edit a file or use a tool, add one line to the permission prompt that says:
1. What it will actually do, in one sentence a 10-year-old would understand. No tech words.
2. A risk level: Safe (only reads or looks at things), Careful (changes or creates files), or Risky (deletes things, overwrites my work, installs software, or sends data off my computer).
3. Exactly what it touches: which files, folders or websites.

Color the risk level green, yellow or red. For anything Risky, also say what could go wrong and whether it can be undone.

Keep it fast, so it never makes me wait more than a second. Name it /explain-commands, so I can type that to turn it on and off.

Idea from command-explainer by Singh-AP.

3. Privacy Guard: stop oversharing with Claude

Before every prompt you send, this checks for personal stuff Claude doesn't actually need: passwords, card numbers, your address, a client's full name, health details, money figures. If it finds any, it stops and asks. One press swaps them for placeholders like [CLIENT_1], and Claude still does the job. Your code is /privacy.

Prompt 3: Privacy Guard
Make me a Claude mod that protects my privacy before every prompt I send.

Before my prompt reaches Claude, check it for personal or sensitive information Claude doesn't actually need to do the job I'm asking for. Look for:
- Passwords, API keys and login codes (always hide these, no matter what)
- Card numbers, bank details, Social Security or ID numbers
- Home addresses, phone numbers and email addresses
- Other people's full names, like clients, coworkers or family
- Health details, money figures, salaries and private business numbers

Then ask one question: does Claude need this to do what I asked? A street address is fine if I'm asking for directions. My SSN is never needed to plan my taxes.

If it doesn't need it, stop the prompt and show me what you found in a bar above the prompt, with the reason for each one, and three buttons:
- Hide it: swap each one for a placeholder like [CLIENT_1] or [ADDRESS_1] and send
- Send anyway
- Let me edit

Do the obvious checks (keys, card numbers, ID numbers) on my computer. Keep a list called my safe list, starting with my own name, that never gets flagged. Keep the real values on my computer only, and put them back if Claude writes them into a file for me.

Name it /privacy, so I can type that to turn it on or off, and /privacy log to see what it caught today.

The key-hiding part builds on redact by karanb192. The "does Claude need this?" check is new.

4 that keep Claude on track

4. Goal Bar: see where you are vs. the goal

Tell it what "done" looks like, and a progress bar follows you the whole chat. It calls out side quests before Claude wanders off. Your code is /goal-bar.

Prompt 4: Goal Bar
Make me a Claude mod that shows me where I'm at in this chat compared to the goal.

At the start of a chat, ask me what "done" looks like in one sentence. Then show a bar above the prompt with: the goal, a progress bar with what percent is done, what Claude is doing right now, and what's waiting on me. Update it after every reply.

If Claude starts working on something that has nothing to do with the goal, flag it as a side quest and ask me if I want to keep going or get back on track. Name it /goal-bar, so I can type that to set or change the goal.

Idea from goal-meter by Nate Herk, where-am-i by hamzafer and aim by Singh-AP.

5. Stop Looping: no more going in circles

When Claude tries the same fix twice and fails twice, this stops it and gives you two other ways forward. Your code is /stop-looping.

Prompt 5: Stop Looping
Make me a Claude mod that notices when Claude is going in circles. If Claude tries the same thing twice and it fails both times, stop it, tell it not to try that a third time, and show me a short note above the prompt saying what it's stuck on and two other ways to fix it. Name it /stop-looping, so I can type that to turn it on and off.

Idea from loop-guard by arasovic.

6. Vague Check: fix the prompt before Claude guesses

"Fix it" gets held, and you get one question plus a clearer version you can send with one press. Your code is /vague-check.

Prompt 6: Vague Check
Make me a Claude mod that catches vague prompts before I send them. If my prompt is too unclear for Claude to get it right the first time (like "fix it" or "make it better"), hold it and show me one short question above the prompt that would make it clear, plus a rewritten version I can send with one press. Name it /vague-check, so I can type that to turn it on and off.

Idea from prompt-bouncer by ray-amjad.

7. Done Check: is it actually done?

Claude sometimes says "done" when it isn't. This checks the claim against what it actually did and flags anything missing. Your code is /done-check.

Prompt 7: Done Check
Make me a Claude mod that checks whether Claude is actually done when it says it's done. Every time Claude says it finished something, look at what it actually did this turn. If anything it said it did isn't true, or it never checked that its work works, show a red note above the prompt with what's missing and a button that tells Claude to finish it. Name it /done-check, so I can type that to turn it on and off.

4 that work around you

8. My Tasks: your own to-do list from Claude

Everything Claude needs from you (log in here, approve this, upload that) lands in one pane you check off. Your code is /my-tasks.

Prompt 8: My Tasks
Make me a Claude mod that gives me my own to-do list for the things only I can do. Whenever Claude needs me to log in somewhere, approve something, upload a file or make a decision, add it to a "My tasks" pane beside the chat instead of burying it in a long reply. Show how many I have left above the prompt, and let me check each one off. Name it /my-tasks, so I can type that to open the pane.

Idea from human-in-the-loop by tzafrir.

9. Help Me Decide: every option, side by side

When Claude asks you to choose, you get a card for each option with the pros, the cons and Claude's pick. Your code is /help-me-decide.

Prompt 9: Help Me Decide
Make me a Claude mod that helps me decide. Whenever Claude ends a reply by asking me to choose between options, open a pane beside the chat that shows each option as a card: what it leads to, the pros, the cons, and which one Claude recommends and why. Let me press a card to pick it. Name it /help-me-decide, so I can type that to turn it on and off.

Idea from decision-tree by ray-amjad.

10. Ding: walk away while it works

A chime and a notification when Claude finishes a long reply or needs your answer. Your code is /ding.

Prompt 10: Ding
Make me a Claude mod that tells me when Claude is done or needs me. Play a short chime and show a notification on my computer when a reply takes longer than 30 seconds to finish, or when Claude is waiting on my answer, so I can walk away while it works. Name it /ding, so I can type that to turn it on and off.

Idea from done-ding by Singh-AP.

11. Recording Mode: safe to screen record

Hides keys, emails, addresses, names and money figures on your screen while you record or share it. Your code is /recording.

Prompt 11: Recording Mode
Make me a Claude mod for when I'm screen recording or sharing my screen. When it's on, hide API keys, email addresses, phone numbers, home addresses, money figures and anyone's full name everywhere on my screen in Claude, and don't let Claude open my .env files or anything with "finance", "invoice" or "contract" in the name. Show a red REC dot above the prompt while it's on. Name it /recording, so I can type that to turn it on and off.

Idea from recording-mode by Nate Herk.

2 just for fun

12. Claude Wrapped: your year with Claude

A Spotify Wrapped-style card of your streaks, peak hours, top projects and your Claude nickname. Your code is /wrapped.

Prompt 12: Claude Wrapped
Make me a Claude mod that makes a Spotify Wrapped-style card of how I use Claude. Show how many days I've used it, my longest streak, the hours I use it most, my top projects, how many things Claude built for me, and a fun nickname for the kind of Claude user I am. Make it look good enough to screenshot and share. Name it /wrapped, so I can type that to open it.

Idea from wrapped by Singh-AP.

13. Arcade: games for the long waits

2048, Tetris and Breakout beside your chat. The game pauses the second Claude is done. Your code is /arcade.

Prompt 13: Arcade
Make me a Claude mod that opens a little arcade beside my chat while Claude works, with 3 simple games: 2048, Tetris and Breakout. Pause the game the second Claude is done and tell me it's my turn. Keep my high scores. Name it /arcade, so I can type that to open it.

Idea from cc-arcade by sezaakgun.

Keep the ones you like

A mod Claude builds this way only lasts for that session, and its folder gets cleaned up later. When you like one, send this in the same chat:

Prompt 14: Keep this mod
Save this mod to a permanent folder and install it for all my projects.

Want more? There are over 800 mods in the ClaudeCodeMod directory, and my other guide has 10 more ways to use Claude mods. Anthropic's mods docs cover the rest.