The argument in one line.
Claude Code's terminal interface hides ten critical productivity features—from planning mode and image analysis to headless scripting—that transform it from a basic coding tool into a full-featured development environment.
Read if. Skip if.
READ IF YOU ARE…
- A developer who uses Claude Code in the terminal regularly and wants to work faster through keyboard shortcuts and lesser-known modes.
- Someone building production applications with Claude Code who needs to understand IDE integration, dual-monitor workflows, and headless scripting for team deployments.
- A developer new to Claude Code who finds the terminal interface intimidating and wants concrete tactical wins to feel more confident immediately.
SKIP IF…
- You're using Claude through the web interface or API only — this breakdown is terminal-native and won't translate to your workflow.
- You're still deciding whether to adopt Claude Code at all — this assumes you're already committed and just wants optimization tips, not foundational justification.
The full version, fast.
Claude Code becomes dramatically more useful once you treat it as a coding partner with hidden modes rather than a plain terminal. The video walks through ten mechanics: shift+tab toggles planning mode so Claude researches and proposes architecture before touching code, Ctrl+V (even on Mac) pastes screenshots for UI analysis, /init writes a CLAUDE.md that auto-loads as context in every future session, IDE integration plus a popped-out window enables two-monitor flow, custom slash commands live as markdown files in .claude/commands, the keyword ultrathink unlocks the largest reasoning budget, double-escape forks the conversation to an earlier prompt, dropping URLs triggers ad-hoc web fetches, and claude -p runs headlessly so you can pipe shell output like npm audit straight into scripted workflows.
Free for members
Chat with this breakdown — free.
Sign in and you get 23 free chat messages on us — ask for the hook, quote a framework, find the exact transcript moment, generate a markdown action plan. Bring your own key when you want unlimited.
Create a free account →Where the time goes.
00:00 – 00:18
01 · Intro / Hook
Claude 4 is the best AI coder, but it lives in a terminal — here are 10 tips to tame it.
00:18 – 01:30
02 · #1 Planning Mode
Shift+Tab twice → planning mode. No code edits, just a written plan with web searches and a project-structure report. Shift+Tab again to leave.
01:30 – 02:46
03 · #2 Image Recognition
On Mac, Cmd+V doesn't paste screenshots into Claude Code — Ctrl+V does. Once pasted, it shows up as 'image 1' and Claude reads UIs surprisingly well.
02:46 – 03:54
04 · #3 /init Command
Run /init to scan the project and generate a CLAUDE.md. Auto-loaded into every new session as context — basically Cursor rules, except free and persistent.
03:54 – 04:29
05 · #4 IDE Integration
Claude Code auto-installs an IDE extension when it detects one (Cursor/VS Code). Two-way comms: highlighting code in the editor scopes Claude's edits; visual diffs appear in the editor.
04:29 – 04:56
06 · #5 Dual Monitor Workflow
Right-click the Claude Code terminal → 'move into new window' → drag to second monitor. Editor on one screen, Claude on the other.
04:56 – 05:59
07 · #6 Custom Slash Commands
Create .claude/commands/<name>.md with arguments and a prompt template. Invoke with /project:<name> <args>. Reusable, project-specific workflows.
05:59 – 06:42
08 · #7 'think' / 'ultrathink' Keywords
Drop 'think', 'think hard', or 'ultrathink' into a prompt to escalate the thinking budget. Gray thinking tokens appear before the answer.
06:42 – 07:30
09 · #8 Escape vs Double-Escape
Esc once = interrupt. Esc twice = jump to any previous message, fork the conversation, and clear history forward. Caveat: auto-accepted edits don't get rolled back.
07:30 – 08:14
10 · #9 Web Search via URL Drop
Outside of planning mode Claude rarely searches the web on its own — but if you paste a URL into the prompt, it'll fetch and use the contents (great for pulling in fresh docs).
08:14 – 09:36
11 · #10 Headless Mode (claude -p)
claude -p '<prompt>' runs Claude in the terminal with no UI, just stdout. Pipe inputs in, pipe outputs to files — e.g. npm audit --json | claude -p '...' > vulnerabilities.md.
09:36 – 09:45
12 · Outro / Subscribe CTA
Hope you learned a thing or two — more AI software-development videos coming, subscribe.
Lines worth screenshotting.
- Headless mode (`claude -p`) lets you pipe Claude into any terminal script, making it a composable piece of any shell workflow rather than just an interactive tool.
- Planning mode does significantly more web searches than normal mode — use it any time you want Claude drawing on current information rather than training data.
- Pasting images into Claude Code on Mac requires Ctrl+V, not Cmd+V — a counterintuitive quirk that blocks most people from using the feature.
- The CLAUDE.md file generated by /init acts as persistent project memory: every future session loads it as context, so switching back after months costs nothing.
- Double-pressing Escape forks the conversation back to any previous prompt, letting you course-correct without starting over entirely.
- Running `npm audit | claude -p` to triage security vulnerabilities by priority is one line of shell script that would have taken a senior engineer an hour to replicate.
- Dropping a URL directly into a Claude Code prompt causes it to fetch and read that page — the fastest way to feed current documentation into your session.
- UltraThink is a literal keyword that increases Claude's reasoning budget and token spend; the gray text you see is its thinking output, not the final answer.
- Custom slash commands stored in .claude/commands/ as markdown files are one of the fastest ways to encode your own workflow patterns into the CLI.
- Claude Code can be split into a separate window and moved to a second monitor, enabling genuine dual-screen vibe coding without terminal hopping.
Steal the format.
The '10 things you didn't know' playbook
Pick a single tool you use every day, find ten quirks nobody talks about, and shoot them as one numbered list with hard cuts and title cards.
- Promise + count in the title ('10 X You'll Wish You Knew Sooner') — high CTR, low ambiguity, evergreen searchable.
- Hard chapter cards (#1, #2, …) double as scrub-bar anchors AND give you ten clean cut points for shorts later.
- Open with the credential ('produces the best AI code I've seen yet') before the friction ('but it lives in a terminal') — that contrast is the hook.
- Hold up the demo app from tip #1 and reuse it through all ten tips. Continuity = comprehension; viewers don't have to rebuild context every tip.
- Wedge the one CTA between two tips, not at the end — viewers who watched 6 tips have proven they care, so the newsletter ask lands softer.
- Skip B-roll. Talking-head + screen recording is enough. The whole thing was probably shot in one hour and edited in three.
- Joe-specific: this format maps perfectly to JoeFlow ('10 JoeFlow tricks…'), Mod Producer ('10 batch-recording quirks…'), and the $6 Stack ('10 things you can self-host this weekend').
Terms worth knowing.
- Planning mode (Claude Code)
- A Claude Code mode (toggled with Shift+Tab) where the AI produces a detailed plan of proposed changes without modifying any code, letting the user review and approve the approach first.
- Headless mode (Claude Code)
- A non-interactive way to run Claude Code via the 'claude -p' command, where a prompt is passed directly and the output is returned to the terminal — useful for scripting and automation pipelines.
- Ultrathink
- A keyword that can be included in a Claude Code prompt to signal the model to spend its maximum thinking budget on a problem, producing more thorough reasoning at higher token cost.
- /init command
- A Claude Code slash command that automatically generates a CLAUDE.md file for the current project by analyzing the codebase, capturing structure, conventions, and setup instructions.
- Custom slash commands
- User-defined shortcuts in Claude Code that trigger a specific pre-written prompt or workflow when typed, allowing repetitive instructions to be stored and reused quickly.
- Pop-out window (Claude Code)
- A detached Claude Code chat window that can be moved independently of the main IDE, enabling dual-monitor setups where code and the AI chat are visible side by side.
- Conversation fork (double-escape)
- A Claude Code interaction where pressing Escape twice abandons the current response and starts a parallel branch from an earlier point in the conversation, preserving the original thread.
- Headless pipeline
- An automated workflow where Claude Code is invoked programmatically by a script or another tool, processing inputs and producing outputs without any human in the loop.
Things they pointed at.
00:38productShopify (used as a sample app the whole video)
04:00productCursor rules (compared to CLAUDE.md)
07:38linkShopify application design guidelines (used as URL-drop demo)
Lines you could clip.
00:23
“If you ever done software development before, you're gonna know that a little bit of planning goes a long way. And Cloud is a secret way to do this.”
01:40
“You'd think on a Mac would just be your normal command V. Click on that, nothing happens. The trick is you actually have to do a control V, which is counterintuitive.”
03:17
“Every time you do a new chat or start a new session with Claude code, it puts this into the context. So if you put this project in the side for six months, came back to Cloud Code later, it would actually bring in all this and remember all this as context.”
06:07
“All the way up to ultra think, which if it sees that word, it's really gonna grind away and try to give you the very best answer.”
06:58
“If you hit the escape key twice, it's gonna give you all the previous messages you had. And then you can go back to the one they interrupted, say. And it's gonna clear the history and go back to that prompt, which is a nice clearing way basically to fork the conversation, get it back on track.”
09:00
“You can do the Claude dash p again for the headless mode, and then ask it to prioritize these security vulnerabilities and describe them what's critical fixes needed, and then pipe all that to vulnerabilities dot m d.”
Word for word.
Read-along
Don't just watch it. Burn it in.
See every word as it's spoken — crank it to 2× and still catch all of it. The same dual-channel trick behind Amazon's Kindle + Audible.
metaphor
00:00Cloud Code powered by the new Cloud four models produces the best AI code I've seen yet. But it does run-in the terminal, which means it can be a bit intimidating at first. There's also a few quirks.
00:10So in this video, I'll show you the top 10 essential tips you need to know to get the most of Cloud Code. Let's go.
00:18So I'm logged in to Cloud Code. And if you ever done software development before, you're gonna know that a little bit of planning goes a long way. And .
00:27So if you go shift tab, you see in the bottom there, it's gonna switch to auto edits, edit on. Do it one more time, and that's gonna go to planning mode on.
00:36And what this mode does is it's not gonna update your code at all. It's just gonna do planning. So let's put in your plan to build a Shopify application that enhances merchants product descriptions with AI.
00:46And in the planning mode, it's gonna start doing some web searches usually, which is great because then your latest information on whatever you're trying to plan to build. Know it does it much more in the planning mode than it does in normal mode, so that's a really good reason to use it.
01:00Then at the end, it gives you a report once in planning mode. So it's gonna tell you the project structure and technology stack to use, what it recommends, core features of this application, implementation steps, all that kind of good stuff.
01:12And then at the end of this game, we'd like to proceed. You can actually start building it, or you can say no, keep planning, and you get it right from there, staying in planning mode.
01:22This is also great for adding a new feature to an application. And then when you're ready to leave, just shift tab again, and you go back to back to normal mode.
01:32You can actually add images directly in the Clog code. It's a really powerful feature, but there's a bit of a trick to it. So let's take this application I have built here, and say we wanna get Clog code to analyze it and maybe add a feature.
01:43So I could just go take a screenshot of it, and now I have a copy to my clipboard. And then in Clog code, you'd think on a Mac would just be your normal command v. Click on that, nothing happens.
01:54The trick is you actually have to do a control v, which is counterintuitive. When I do that, now image one comes up.
02:03So it doesn't give you a preview or anything, but when you see, like, image number one, that's the image you just copied from your clipboard. I haven't tried it on Windows yet. It might be the same kind of trick.
02:11Maybe it's the Windows key or something, but watch out for that. But once you do get it in here, it's very powerful. It's just, like, do an analysis of this application's UI, and then what enhancements would you recommend?
02:21And I find this very useful for just looking at bugs in a UI or just adding a feature, adding a button somewhere. It actually is very good at reading the image. And there we go.
02:30It gives us some functionality enhancements. These are a couple of things I actually thinking about into the application, so it's good that it picked this up, like directions integration and store hours, stuff like that. Definitely check out how Cloud Code can read images.
02:42It does a really good job.
02:46One really powerful command that I think is a bit misunderstood and overlooked is the init command. So if you're in your project and you go to init, what it's gonna do is look through all the files in that project and write a claud dot m d file or markdown file.
03:02And it'll give you a comprehensive look at your whole project. It'll give you coding standards, technologies used, etcetera.
03:08And then we see it create the claw dot m d file in this directory. And then, like I said, it gives you the core components, data flow, all that kind of good stuff. And I've been doing software development for a long time.
03:18This is some of the best kinda lightweight documentation you can get, especially for a smaller code base. But the really interesting thing is it takes this Claude dot MD file, and every time you do a new chat or start a new session with Claude code, it puts this into the context.
03:33So if you put this project in the side for, say, six months, came back to Cloud Code later, it would actually bring in all this and remember all this as context, which is a nice head start.
03:43But, also, you can add on to this yourself. So if there's things you want Cloud Code to know when it launches up for this project, this is the place to do it.
03:51This is kinda like cursor rules, basically.
03:55If you're used to doing your coding in a code editor, one thing that will really help is if you integrate a code editor with Cloud Code, so you're not always in the terminal. It's really easy to do once you install Cloud Code and it detects you have an IDE, it actually automatically installs it.
04:08There's also an IDE command you can run, and that'll tell you what IDs you have and connect to it from there. Once you have it connected, it's gonna tell you what file you're in.
04:17And then it has two way communication. So if I highlight some code, it's gonna say 12 lines selected.
04:22So it's just targeting that code to make changes on. It also lets you visually see the differences between a code change you're making.
04:31So once you have Cloud Code hooked up with your IDE, in this case, have Cursor, the next tip is to actually keep Cloud Code and Cursor or Versus Code, whatever you're using for your IDE, in a separate window. It's gonna make it such a nice experience to code in.
04:45As to do that, all you have to do is just when you have Cloud Code up, you can just right click on that terminal and say move into new window. Now you just move that to your second monitor and really do some vibe coding.
04:56As well as using the built in commands in Cloud Code, you can actually create your own and they can do whatever you want and they're very useful. So to do that, we'll create a new directory. So you do do a make directory under dot Cloud and then you call it commands.
05:08And then you create the commands by just simply creating markdown files. That's what I'm gonna call doc dot m d. And then inside of it, you can put your arguments.
05:17So I'm gonna say create clear thorough documentation for the arguments. That's gonna be the files we pass in. Then include a description, usage examples, and important notes for developers.
05:27And once that's created, you can just do slash project colon and then the name of the command, so in this case doc. And then passing your arguments, so we'll pass in store locator dot html.
05:36So now it's reading that input parameters. You can make many arguments for whatever you think you can make up in terms of a command that's useful to you. So it's gonna read through that.
05:46And then it just gives you what you asked for basically in the prompt. So this case is a really nice piece of documentation describing this particular file.
05:54I mean, the sky's the limit what we can do with these custom commands, so they're gonna save you a lot of time.
06:00Cloud Code actually has some keywords that tell it how much to think, and they all have the word of think in them. So if you say think, it's actually gonna reason more, spend more time, chew more tokens, all the way up to ultra think, which if it sees that word, it's and try to give you the very best answer.
06:18Let's try out let's just say UltraThink, the best way to package this Shopify application, then merchants can deploy it. And then it came back with the answer.
06:25But you'll notice in this case, the gray text is all the thinking tokens it used along the way. Definitely very useful to know in certain situations. If you're into AI software development like me, make sure you subscribe to my newsletter, the AI unleash news.
06:38It's the first link in the description, and I hope to see you there.
06:43I think most people know that if cloud's going in the wrong direction, like, it's not really following the prompt or you can just see right away it's kinda doing the wrong thing. If you just hit the escape key in the middle, then anytime you're gonna interrupt it.
06:56But what you can also do, which is very useful, is if you hit the escape key twice, it's gonna give you all the previous messages you had. And then you can go back to the one they interrupted, say.
07:07And it's gonna clear the history and go back to that prompt, which is a nice clearing way basically to , get it back on track to what you actually want it to be. One thing to keep in mind though is if you do do auto edits and it and they were accepted already, and then you just press escape key and then you go back, it's not gonna revoke those and go back to this breakpoint, but it will take the conversation to this point.
07:31Cloud Code can look at websites and do web searches. It just doesn't do it very often, I've noticed. Except for that is when you go into planning mode.
07:38But say you're just doing coding. You can actually put in URLs and it will go and look at them. So I'm say make this application conform to these style guides.
07:46Then I'm gonna paste in the Shopify application design guidelines. And now it's gonna go out and actually fetch all the content from that website. So this is a great way to just bring in documentation, the latest documentation you have into Cloud Code.
07:59But now you see it's gone out to that website, come back with very specific things to do to make it conform with the style guidelines. So, for example, updating the typography to match the Shopify admin interface.
08:09So it's really got all the great information we need here. Now it's gonna update the code.
08:15Alright. We're down to the last one and I find this one so useful. I use it all the time.
08:19You can actually run Claude code in headless mode, which means it doesn't actually load up the UI or anything. So check it out. You can just say clod in any terminal.
08:28You say clod dash p for headless. That's the headless mode switch. Let's ask the question, what is the capital of Slovakia?
08:34It's just going to return the response to me in the command line. So what this means, we can use it all kinds of different places, pipe it to different files, use it to look at images, all kinds of useful functionality just by using the claw dash p command.
08:47So check out this much more useful example. If I run an NPM audit, which is gonna tell me all the vulnerabilities with all the packers I have in my application, then take the results of that, pass them to Claude, so I can do the Claude dash p again for the headless mode, and then ask it to prioritize these security vulnerabilities and describe them what's critical fixes needed, and then pipe all that to vulnerabilities dot m d, which will a markdown file which will have all that information.
09:11Now it generated this vulnerabilities file, which gives me all the critical things I need to do, like, right away. For example, this one is a Next. J s authorization bypass vulnerability.
09:20So I have to get that figured out right away. And then it lists the high priority and low priority from there. So I'm sure your head's spinning all the different places you could use Clog code in headless mode.
09:28It's such a powerful tool that add any script, any kind of workflow. I hope you learned a thing or two.
09:34Clock code is great, and I'm do lots of videos on it coming up, as well as everything else around AI software development. So if you're into that, make sure you subscribe to the channel. And I hope you have an amazing day.
09:43I'll talk to you in the next one.
The Hook
The bait, then the rug-pull.
The whole video is a single, well-disguised promise: "Claude Code is the best coding model out there, but it's hiding ten of its best tricks behind quirks and keyboard shortcuts no one told you about." AI Unleashed opens with the Claude Code splash screen on one side and his face on the other — terminal-shy viewer, meet patient teacher — and immediately launches into tip #1 before anyone can scroll away.
Named ideas worth stealing.
00:28concept
Shift+Tab mode cycle
Shift+Tab once = auto-edits on; twice = planning mode on; third time = back to normal. One key, three workflow modes.
Steal forAny agent tool with multiple operating modes — overload one key with a visible state line at the bottom of the terminal.
02:48concept
Project memory via /init → CLAUDE.md
/init scans the codebase and writes a markdown file (architecture, conventions, components). Auto-loaded into every new session. You can hand-edit it to inject permanent context.
Steal forPersistent agent memory pattern — one auto-generated, hand-editable file that every session reads on boot. Same trick works for any LLM wrapper Joe builds.
05:59list
Thinking-budget keywords
- think
- think hard
- think harder
- ultrathink
Magic words inside a prompt that escalate Claude's reasoning budget. The gray text streamed before the answer is the thinking trace.
Steal forSurface this as a user-facing knob in JoeFlow / Mod Producer — give Joe a dial that says 'normal / hard / ultra' that prepends the keyword automatically.
04:56concept
Custom slash commands as .md files
Drop a markdown file at .claude/commands/<name>.md with $ARGUMENTS placeholders → invoke as /project:<name> <args>. No code, no plugin system, just files.
Steal forTemplated batch prompts — ship JoeFlow with a /commands folder so users can swipe Joe's prompt templates (Killing Excuses, Sip Ship Sell, doc-gen) like recipes.
08:15concept
Headless mode as a Unix citizen (claude -p)
claude -p turns Claude Code into a stdin → stdout filter. Composes with pipes, redirects, and any other CLI tool.
Steal forBuild CLI wrappers that pipe data through Claude as a step in a normal shell pipeline — e.g. transcript → claude -p 'extract quotables' > quotables.md. This is the unlock for Joe's morning batch launcher idea.
How they asked for the click.
VERBAL ASK
09:36subscribe
“Make sure you subscribe to the channel. And I hope you have an amazing day. I'll talk to you in the next one.”
Soft outro CTA, no urgency. Mid-roll newsletter pitch at 6:30 is more interesting — wedged between two tips so it doesn't feel like a sponsor break.
MENTIONED ON CAMERA
FROM THE DESCRIPTION
PRIMARY CTAWhere the creator wants you to go next.
Visual structure at a glance.
