E1 · Two commands
10 minIn a session for one of your projects, type /output-style and choose concise. Work for a bit, then type /recap.
Done when: The replies got shorter, and /recap gave you a one-line summary of what you did.
Lesson 7 · 16 min
Five short commands change how Claude works: undo, see what’s taking up space in the conversation, summarize, work toward a goal, and give shorter answers. Mods change how Claude Code looks, but you should check them first. At the end, Carla picks three new things from the course and writes her routine for the week. You’ll make yours.
Start the lesson
Goal: Adjust Claude with slash commands and mods, then build your routine with the new things you learned.
Tap or click each part. A routine is what turns a tool into a habit.
Scroll down: each step below lights up its part of the diagram.
A slash command is a shortcut that starts with / and changes how Claude works in that session. Type / in the message box and choose from the list. No coding needed.
A mod is a type of plugin (add-on), which is why its menu is called /plugin. It changes how Claude Code looks and behaves: a bar showing how much of the conversation you’ve used, an alert when something finishes, or a new command. It appears in the app’s Code tab (and also in the terminal and VS Code, a code editor).
A mod doesn’t run on its own: it can access whatever Claude Code can, including your files. So before installing, ask Claude to check what the mod does. If something goes wrong later, ask Claude to turn off the mod or use the /plugin menu. This part is optional and more technical.
No one uses thirteen new things all at once. Pick three, say what day or when you’ll use each one, and why. A week from now, replace the one that didn’t stick.
Commands for everyday use, mods only after checking them, and a short routine with three new things.
Commands adjust the session; mods adjust Claude Code; your routine decides what you actually use.
Type / in the message box (in the app’s Code tab) and choose from the list. If a command doesn’t appear, update the app and check again. If it still doesn’t appear, ask Claude: "how do I use /recap in this app?". In the terminal (optional), they work the same way.
Takes the conversation back to before a bad answer, as if it never happened. The shortcut is to press Esc twice.
Better than arguing with Claude: go back and ask in a different way. It rewinds the conversation; if the bad answer already changed a file, ask Claude to check the file.
She asked only for the title of lesson 1, and Claude replied with the whole lesson. /rewind, then a new, clearer request.
Context is everything Claude is considering in this conversation: history, files it has read, and instructions. /context shows a color-coded chart of how much space each part takes up.
A full conversation gets slower and uses more of your limit.
She saw that three old scripts were taking up a lot of space, so she started a new session for lesson 2.
Gives you a one-line summary of the session right away. Useful when you come back from lunch and can’t remember where you left off.
After a few exchanges, Claude also makes this summary on its own.
Give Claude a clear goal, and it keeps working until it reaches it. A clear goal can be measured: "5 exercises, each with an answer," not "good exercises."
Without a measure, it doesn’t know when to stop.
Changes how the responses are written. Choose concise for shorter, more direct answers.
Good for everyday use; switch back to the default style when you want a longer explanation.
One at a time, in the course folder’s session.
/output-style concise /recap
/goal Create curso-planilhas/aula-2/exercicios.md with 5 VLOOKUP exercises, each with a sample spreadsheet and the correct answer.
/context
Lesson 2 outline (VLOOKUP) written and approved; the exercises are still missing.
Goal reached.
You can skip this if you just want to use the app as it is. Ready-to-use examples: 18 mods in Portuguese, tested, in the inematds/inema-mods repository on GitHub. GitHub is a site where people store and share code; a repository is a project folder there.
Mods require Claude Code version 2.1.287 or later. In the app, just keep it up to date; in the terminal (optional), claude --version shows the number.
Older version: the mod won’t appear, and it’s not a problem with the mod.
Mods aren’t isolated: they can read your files and run commands, just like Claude Code. So have Claude check first: it runs a check command (claude plugin validate) in the mod’s folder.
This check shows when the mod acts on its own (the hooks) and what it asks to use. You don’t need to understand the command: the prompt below asks for a summary in plain language.
In the app’s Code tab, paste the prompt below: Claude will run the check itself.
Read the plain-language summary: what it reads, what it runs, and whether it uses the internet. Unsure? Don’t install it.
To install or turn off mods, type /plugin in the message box and use the menu that opens. The change takes effect when you close the menu.
If Claude starts acting strangely after you add a mod, ask in that session: "turn off the mod [name]" or "uninstall the mod [name]". Or type /plugin in the message box and turn off the mod in the menu; it takes effect when you close the menu.
For terminal users only (optional): claude --safe-mode opens Claude Code without any customizations if it won’t open properly.
Paste this into a session in the Code tab or terminal. Replace the link with the link to the mod you want.
I want to check a mod before installing it: <link to the mod on GitHub> Examples of mods: https://github.com/inematds/inema-mods Download it to a separate folder and run claude plugin validate there. Explain it in up to 5 lines, without technical terms: - when it acts (hooks) - whether it reads my files, runs commands, or uses the internet - how to turn it off later Don’t install anything until I reply YES.
Turn off the mod <mod name> and tell me what the list of enabled mods looks like.
claude --safe-mode
What this mod does: shows a bar in the footer with how much of the conversation has been used.
Validation passed without errors. Should I install it? Reply YES.
If any of these apply, don’t install it.
Write 3 numbered lines: new thing | when you use it | what for. The page checks whether each line mentions something new from the course and a day or time, and whether the 3 new things are different. This is a format check: you decide whether the routine works for you.
You can manage commands and routines in the app. Checking a mod (E2) is optional and more technical.
In a session for one of your projects, type /output-style and choose concise. Work for a bit, then type /recap.
Done when: The replies got shorter, and /recap gave you a one-line summary of what you did.
Choose a mod from inematds/inema-mods and paste Carla’s prompt with its link. Read the summary before replying.
Done when: You have, in up to 5 lines and in simple words, what the mod can access and how to turn it off; installing it or not is your choice.
Choose 3 course features you’ll use and write one line for each: feature | when you’ll use it | what for. Paste them into the checker above and save as rotina.md: the checker’s "Download file" button saves the ready-to-use file. Or ask Claude, in the Code tab: "create rotina.md in the project folder with this text" and paste in your routine.
Done when: rotina.md is saved, and the checker replies "OK: routine with 3 features, each with a day or time".
The explanation appears after your answer.
Question 1 of 5
Claude gave you a bad answer, and you want to ask another way as if it hadn’t happened. What’s the fastest way?
Question 2 of 5
Why ask Claude to check a mod before installing it?
Question 3 of 5
Which of these is a good /goal?
Question 4 of 5
You installed a mod, and Claude Code started acting strangely. What should you do first?
Question 5 of 5
Which of these is a good routine?
Choose an option. The explanation appears after your attempt.