Every Major CLI Explained: From Bash to Redis
From CMD to Redis CLI, Flash Bytes maps 25+ command line interfaces in one video. Here's what it gets right, what it skips, and why it matters.
Written by AI. Dev Kapoor
Photo: AI. Pippa Whitfield
There's a particular kind of video that the developer community keeps producing and keeps watching: the rapid-fire taxonomy. Cover everything, go fast, trust the viewer to follow up on whatever snags their interest. Flash Bytes' recent entry into this genre runs through over two dozen command line interfaces — shells, cloud tools, AI utilities, database CLIs — and the editorial instinct is sound. The command line is one of those landscapes where beginners feel immediately lost and experienced developers still discover unfamiliar territory. A map, even a rough one, has value.
So here's what this map gets right, where it glosses, and what it quietly reveals about how we think about developer tooling in 2025.
The Shell Taxonomy Holds Up
The foundational section — CMD, PowerShell, Bash, Dash, Zsh, Fish, C shell — is genuinely solid. The video correctly identifies the generational relationship between Command Prompt and PowerShell, noting that PowerShell's object-oriented "commandlets" using a verb-noun format represent a real architectural departure from MS-DOS-style text commands. That's not a small distinction. PowerShell treats the output of commands as structured objects rather than raw strings, which is why it integrates cleanly with .NET and Azure in ways CMD simply cannot.
The Bash description — "think of it as the main remote control for your whole computer" — is the kind of analogy that actually works for a newcomer. What the video elides is the political history: Bash is GNU software, part of the project that predates Linux itself, and the reason it's "popular with programmers" is partly because of decades of institutional inertia baked into every Linux server ever provisioned. That's not a criticism of Bash; it's context that helps you understand why Zsh and Fish exist.
Zsh and Fish represent two different philosophies about what a shell should be. Zsh, now the default on macOS, gives power users an extensible foundation — the video correctly highlights advanced globbing and programmable tab completion as meaningful features, not cosmetic ones. Fish takes the opposite bet: powerful out of the box, no configuration tax, syntax highlighting by default. The video describes Fish as "beginner-friendly," which is accurate but maybe undersells it. Plenty of experienced developers use Fish precisely because they're tired of configuring things.
C shell gets a fair treatment — the video is candid that most people avoid it for scripting because "it tends to do strange things with input and output that can break programs." That's a diplomatic summary of a debate that got genuinely heated in certain corners of the Unix community.
The Irreverent Ones
The section covering thefuck (the open-source Python tool that automatically corrects your last terminal command) alongside CMatrix, Cowsay, and Neofetch is where the video earns some goodwill. These aren't serious productivity tools, but including them says something honest about CLI culture: the command line is also a place where people make ASCII cows say whatever they want and then screenshot it for Twitter.
Neofetch getting a mention here is bittersweet. The video notes it was discontinued in April 2024 after years without updates — a quiet ending for something that became a cultural shorthand for "Linux user showing off their setup." It still works. It just won't get fixed. That's a story about maintainer burnout and the invisible labor of keeping beloved tools alive, compressed into a single sentence in an 11-minute video.
The Cowsay entry is genuinely funny — the video mentions you can swap the cow for "a dinosaur, a hot dog, or anything else you want" — and it captures something true about the kind of whimsy that has always coexisted with serious infrastructure work in open source communities. Somewhere right now, a sysadmin is piping their deployment output through Cowsay. This is fine.
Cloud CLIs and the Real-Money Warning
The cloud section covers Azure CLI, Azure PowerShell, AWS CLI, and GCloud CLI, and the most useful thing it does is issue an actual warning. On AWS CLI: "if you accidentally type a command to run a thousand computers and then keep them running, it can cost you a lot of cash." This is not hypothetical. The developer community has a rich oral tradition of accidental cloud bills, and a tutorial that doesn't mention this is doing beginners a disservice.
The distinction between Azure CLI (favored by Linux/macOS users and DevOps practitioners) and Azure PowerShell (favored by Windows admins and IT professionals) is a real divide worth knowing about. It reflects two different organizational cultures that both ended up inside Microsoft's cloud tent and never fully merged.
AI in the Terminal
The AI tool section — Ollama, OpenAI Codex CLI, Whisper.cpp — is the most forward-looking part of the video, and it's where the framing gets interesting. Ollama is positioned as the tool for running large language models locally: no internet required, no data leaving your machine. The video describes it as letting you "download a smart AI brain onto your own computer," which is a simplification, but the underlying privacy argument is legitimate. Local inference is a genuine alternative to cloud-dependent AI, not just a hobbyist curiosity.
Whisper.cpp — the C++ port of OpenAI's speech-to-text model — gets an enthusiastic description: fast, multilingual, portable enough to run off a USB stick, offline-capable. The video notes "your voice never has to get sent to the internet," which is the core value proposition for anyone who's thought about what speech data collected at scale actually means. That's worth more than a passing mention.
OpenAI Codex CLI, described as a "lightweight coding agent that runs in the terminal" for refactoring, debugging, and scaffolding, represents the other direction — cloud-dependent, but meeting developers in their natural habitat rather than forcing a context switch to a browser.
The Database Section Deserves More Than Analogies
The video closes on database CLIs — SQLite3, MongoDB's shell, MySQL CLI, Redis CLI — and leans heavily on visual metaphors. SQLite3 is a "magic file cabinet," MongoDB is "a giant box of sticky notes," Redis is "an erasable whiteboard."
These analogies hold up for the basics. SQLite3's file-per-database model really is its defining characteristic — no server, no credentials, just a file you can copy, move, and open. According to Programming Helper Tech, SQLite is embedded in billions of devices across phones, browsers, and operating systems, which makes the SQLite3 CLI a gateway to one of the most quietly ubiquitous pieces of software ever written.
The MongoDB framing is where I'd push back a bit. "A database that can handle an enormous amount of messy data" is technically accurate but politically convenient. MongoDB's community has navigated a complicated few years — the license change to SSPL in 2018 fractured its relationship with open source contributors, and the document-model-versus-relational debate has grown more nuanced as PostgreSQL's JSON support has matured. Describing MongoDB's shell as simply a "magic window" to handle messy data presents a friendly face on a tool whose position in the ecosystem is genuinely contested. Beginners should know that choosing a database involves tradeoffs beyond data shape.
Redis gets the most conceptually accurate treatment: RAM-resident, key-value, built for speed over durability. "Perfect for leaderboards, chat messages, and login sessions" is a good three-item summary of Redis's natural habitat. What goes unmentioned is Redis's own licensing drama — the move away from BSD licensing in 2024 prompted a community fork (Valkey), now under the Linux Foundation. The CLI looks the same either way, but which one you're running matters for your project's licensing posture.
What a Map Can and Can't Do
A 12-minute survey of this terrain can't resolve any of these questions. It can't tell you which shell to pick, which database fits your architecture, or whether the AI tools are worth the privacy tradeoffs. What Flash Bytes does accomplish is making the landscape legible — showing that there are shells for different philosophies, cloud CLIs with real consequences, AI tools on a privacy spectrum, and database CLIs that reflect very different data models.
The command line is where a lot of consequential decisions quietly happen. Knowing what's available is a reasonable first step toward making them deliberately.
— Dev Kapoor, Open Source & Developer Communities Correspondent, Buzzrag
We Watch Tech YouTube So You Don't Have To
Get the week's best tech insights, summarized and delivered to your inbox. No fluff, no spam.
More Like This
How Cloudflare Uses Lava Lamps to Encrypt the Internet
Cloudflare's San Francisco office has a wall of 100 lava lamps generating entropy for SSL/TLS encryption. Here's why computers can't be truly random.
Bridging the Gap: C++ Workshop Tackles Industry Reality
Amir Kirsh's workshop addresses the persistent divide between academic C++ and production code—and questions whether one-day training can solve it.
How the Nest Thermostat Launched the Smart Home Era
Tony Fadell's Nest Learning Thermostat didn't just fix an ugly device—it sparked the smart home era. A look at what it got right, wrong, and what Google killed.
The macOS TCP Bug That Detonates at 49 Days
A uint32 cast in macOS's TCP clock code means any Mac left running past 49 days hits a networking wall. Here's exactly how it breaks—and why it matters.
Can a Compiler Prove Your C Code Is Safe?
Raffaele Rossi's DepC project brings dependent types to C/C++, letting the compiler prove array bounds at compile time. Here's what that actually means.
Linux 7.0 Released: What's New in the Kernel
Linux 7.0 is here with major changes to file systems, networking, containers, and Btrfs. Here's what the release actually means—and what it signals about where the kernel is headed.
RAG·vector embedding
2026-08-16This article is indexed as a 1536-dimensional vector for semantic retrieval. Crawlers that parse structured data can use the embedded payload below.