free · no signup · 112 entries and growing
The Vibe Coder's Dictionary
Your AI speaks fluent developer. You don't have to. But understanding its words is how you stay in charge, so each entry explains one term the way a friend would, shows why your agent brought it up, and teaches you how to say it like a dev.
AI & agents
Context windows, tokens, MCP, hooks: the machinery behind your agent.
Claude Code vs Claude.ai: which Claude are you actually using?
Claude.ai is the chat where you paste code and read answers, while Claude Code is the agent that opens your project, edits real files, and runs commands, all powered by the same Claude models.
Cursor modes explained: Agent, Ask, and where Composer went
Cursor's chat pane runs in modes that set what the AI may do: Agent edits your files, Ask answers questions read-only, and Plan, Debug, and Design cover special workflows; Composer, the old pane name, now names Cursor's own model.
Do I need an AGENTS.md? (and how it differs from CLAUDE.md)
AGENTS.md is an open, plain markdown file at your project's root that gives AI coding agents your build and test commands, conventions, and warnings in one shared format, read natively by tools like Cursor, Codex, and GitHub Copilot, though Claude Code reads CLAUDE.md instead.
Lovable Chat mode vs default mode: when to use which
Chat mode, renamed Plan mode, is Lovable's talk-only mode for planning and debugging without touching code, while the default Build mode is the one that edits your project.
Lovable credits: what actually burns them (and how to burn fewer)
Lovable credits are the currency Lovable charges for AI work: every Plan mode message costs a flat 1 credit, and every Build mode request costs a variable amount based on how much code the agent reads, changes, and verifies.
The one-feature rule: why smaller asks get you better code
A working habit for building with AI: describe one feature, let your agent build it, test it yourself, commit it, and only then start the next one.
Tokens vs credits: what you are actually paying for when you vibe code
Tokens are the raw unit AI models bill by; credits are the wrapper currency your coding tool converts them into, at an exchange rate that isn't fixed.
Usage limits in Claude Code (windows, caps, and how not to hit them)
Usage limits are the two stacked caps on your Claude plan, a five-hour rolling session window and a separate weekly ceiling, both measured in tokens rather than message count, that block further use until they reset.
What are Claude skills? (reusable instructions your agent loads on demand)
A Claude skill is a folder with a SKILL.md file of instructions, and optionally scripts or templates, that Claude Code reads and follows only when your request matches what the skill is meant to do, instead of every procedure sitting loaded all the time.
What are hooks in Claude Code?
A hook is a command you wire to a specific moment in Claude Code's process (like after a file edit or when it finishes a turn) so it runs automatically without you asking.
What are subagents in Claude Code?
A subagent is a separate Claude instance your coding agent spawns to handle one bounded task in its own context window, then hands back just the result.
What is /compact in Claude Code? (and why it can lose your context)
/compact is Claude Code rewriting your conversation into a shorter summary when the context window is filling up, and summaries always drop detail.
What is a context window? Why your AI forgets what you were doing
A context window is the fixed amount of conversation and code your AI can actually see at once. Once it's full, older stuff quietly falls off.
What is a model router, and what does 'Auto' mode do?
A model router is the system inside an AI tool that automatically decides, request by request, which underlying AI model answers you, trading capability against cost and speed instead of using one fixed model for everything.
What is a system prompt? (and why the same AI feels different in every tool)
A system prompt is the set of instructions an app hands the AI before you type your first word, and it's why the same underlying model can feel like a different assistant in every tool you use it in.
What is agentic coding? (and how it differs from vibe coding)
Agentic coding is when an AI works through a task in a loop, reading your files, taking real actions, checking the results, and adjusting, instead of answering your prompt once and stopping.
What is AI slop? (correct syntax, wrong everything)
AI slop is code that looks finished and correct, clean syntax, sensible structure, but doesn't actually do what it's supposed to once you test it.
What is an AI hallucination in code? When your agent invents things
An AI hallucination in code is when your agent confidently writes a function, library, or setting that simply doesn't exist.
What is context engineering? (and is prompt engineering dead?)
Context engineering is the practice of deliberately curating everything that loads into a model's context window before it answers: which files, which past messages, which tool results, and which instructions, so the model has what it needs and nothing crowding it out.
What is function calling? (how your AI actually uses tools)
Function calling is the mechanism where an AI model outputs a structured request naming a tool and its arguments, and the app around it executes that request and returns the result, letting agents act instead of just talk.
What is MCP (Model Context Protocol)? And do you actually need it?
MCP is the open standard that lets your AI agent plug into outside tools and data (like GitHub, Notion, or a database) without custom code for every single one.
What is Plan Mode in Claude Code? (and what it does not protect you from)
Plan Mode blocks Claude Code's file-editing tools until you approve a plan, but a shell-command prompt you approve during that time still runs for real.
What is prompt fatigue? (when fixing one bug takes fourteen prompts)
Prompt fatigue is the exhausting spiral where you send prompt after prompt trying to fix one bug, and each fix patches the last patch instead of closing in on the actual cause.
What is RAG? (how your AI looks things up before answering)
RAG (Retrieval-Augmented Generation) is when your AI searches a knowledge base for relevant information first, then writes its answer using what it found, instead of answering purely from memory.
What is the 70% wall? (why the last 30% of your app is the hard part)
The 70% wall is the point in building an app, usually hour four to eight of a first project, where sending another prompt stops closing the gap between mostly working and actually working.
What is vibe coding? (the name for what you are doing)
Vibe coding is building software by describing what you want to an AI in plain language and letting it write the code, originally with the twist that you never read the code at all.
What is YOLO mode, and should you skip permissions?
YOLO mode is the common nickname for any AI coding agent setting, such as Claude Code's bypassPermissions mode or Cursor's Run Everything mode, that lets the agent edit files, run shell commands, and make network requests without pausing for your approval.
Code concepts
Components, state, frameworks: the ideas your agent builds with.
Framework vs library: what is the difference?
A library is code your app calls when it needs something; a framework calls your code and decides the structure, which is why your whole project takes its shape from the framework you pick.
Frontend vs backend: which half of your app is which?
The frontend is the half of your app that runs in each visitor's browser, everything they see and click; the backend runs on a server you control, where secrets stay hidden and the real rules get enforced.
Local storage, session storage, cookies: where your app remembers you
Cookies, localStorage, and sessionStorage are the three places a browser stores data for your app: cookies ride along to the server and can expire, localStorage stays until cleared, sessionStorage dies with the tab.
What are props (and how are they different from state)?
Props are the labeled, read-only inputs a parent component passes down to a child component, flowing one way and never changed by the child that receives them.
What does "Module not found: Can't resolve" mean?
Import and export are the JavaScript keywords that let one file hand off a function, variable, or component to another file, and a mismatched path or export style is exactly what "Module not found" and "Cannot find module" errors are reporting.
What is a component? (the building blocks your AI keeps creating)
A component is a self-contained piece of your app's interface, like a button or a navbar, that bundles what it looks like with how it behaves, so it can be defined once and reused anywhere.
What is a feature flag, and how do you test on just you?
A feature flag is a switch stored outside your code, often an environment variable or a dashboard toggle, that turns a feature on or off without a new deploy.
What is a function? (and why your AI keeps extracting them)
A function is a named block of code that takes input values called parameters, does some work with them, and hands back a result through a return statement, so that work can be reused anywhere just by calling the function's name.
What is a hydration error in a Next.js or React app?
A hydration error is React's warning that the interactive version it just built in the browser doesn't match the plain HTML the server sent down, usually because some value rendered differently on each side.
What is a React error boundary (and why did my page go white)?
An error boundary is a piece of a React app that catches a crash in the components rendered inside it and shows a fallback message instead of leaving the whole page blank.
What is a variable? (the names your code remembers things by)
A variable is a name your program attaches to a value, like a form input or a running total, so later lines of code can use and change that value by name.
What is an edge case? (why your app breaks on weird inputs)
An edge case is input or timing at the boundary of what your code expects (an empty field, a giant file, a double click), the stuff that breaks apps that worked fine in the demo.
What is async/await and why is my data undefined?
Async/await is JavaScript syntax where the async keyword marks a function as one that returns a promise and await pauses that function until a called promise settles into its real value, so asynchronous code reads top to bottom instead of chaining callbacks.
What is CRUD? (create, read, update, delete)
CRUD stands for create, read, update, delete: the four basic operations an app performs on stored data, and the pattern almost every feature is built from.
What is debounce and why does your search box wait?
Debounce is a technique that delays running a function until a burst of rapid calls, like keystrokes, goes quiet for a set pause, so it runs once instead of on every single event.
What is JSON? (the format your tools speak to each other)
JSON is a plain-text format for structured data: curly braces hold named fields, square brackets hold lists, and nearly everything in your stack uses it, from package.json to API responses.
What is middleware? (the code that runs between request and response)
Middleware is code that runs between an incoming request and the page or endpoint that answers it, where your app checks logins, redirects visitors, logs traffic, or blocks requests before they reach your routes.
What is refactoring? (changing code without changing what it does)
Refactoring means restructuring existing code so it's easier to read and change while keeping its observable behavior exactly the same: same features, same outputs, nothing new added.
What is routing? (how URLs become pages)
Routing is how your app matches a URL to the page or endpoint that should answer it; when no route matches, the visitor gets a 404.
What is state? (why your app remembers, and why it forgets)
State is the information your app holds in memory while it runs (who is logged in, what is in the cart, what you typed), and the screen you see is drawn from it.
What is technical debt? (the interest your shortcuts charge)
Technical debt is the future cost of today's shortcuts: every quick fix and duplicated block makes the next feature slower to build and easier to break, like interest on a loan.
What is TypeScript? (and why your AI writes it by default)
TypeScript is JavaScript plus type annotations that describe what shape your data has; the compiler checks those shapes and converts everything to plain JavaScript, catching mistakes at build time instead of in front of your users.
What's the difference between undefined and null?
undefined means a variable, property, or argument was never given a value, while null means something was deliberately set to empty, and JavaScript throws a TypeError the moment your code tries to read a property or call a method on either one.
Why does my like button flip back after I tap it?
Optimistic UI is a pattern where the screen updates instantly as if an action already succeeded, then quietly reverts if the server request behind it actually fails.
Git & versions
Commits, branches, worktrees: how your project remembers.
Git clone vs pull vs fetch: which one do you need?
Git clone downloads a full copy of a repository the first time you get it, git fetch checks the remote for new commits without touching your files, and git pull fetches those commits and merges them straight into your current branch.
How do you undo in vibe coding? Rollback, revert and version history
Rollback, revert, and version history are different ways of getting your project back to a state that worked before the last change broke it.
What is a git worktree? (and why your coding agent created one)
A git worktree is a separate folder checked out from your same project history, letting your AI agent work on one thing there while your main session keeps working untouched elsewhere.
What is a merge conflict? (two changes, same lines, someone has to choose)
A merge conflict is what git shows you when two sets of changes edit the exact same lines of a file in different ways, and it needs a human to pick which version survives.
What is a pull request? (proposing changes instead of pushing them)
A pull request is a request to merge one branch of code into another, packaged with a diff and a review step, instead of pushing changes straight onto the main branch.
What is Git? Commits, branches and pushes, explained for vibe coders
Git is a save-and-rewind system that records labeled snapshots of your project (commits), lets you branch off to try things safely, and pushes those snapshots to a remote copy like GitHub.
Terminal
The black window, decoded.
npm says I have vulnerabilities. Am I hacked?
npm vulnerabilities are known security flaws that npm audit finds by checking your installed packages against the GitHub Advisory Database, then reports as a count sorted by severity, from low to critical.
npm vs npx: which one do you type, and why?
npm installs and manages the packages your project keeps; npx runs a package's command once, using your project's copy if it exists or downloading one to npm's cache if it doesn't.
Permission denied: what your terminal is actually saying (and the fix you should never copy-paste)
"Permission denied" is your operating system telling you that your user account isn't allowed to do what you just asked, not that something is broken.
What are sudo and chmod? (and when it is safe to run them)
sudo runs a single command with administrator power, and chmod changes who may read, write, or run a file; both are common fixes for permission errors, and both deserve a look before you press Enter.
What do npm install, node_modules and package.json actually do?
npm install reads package.json's list of dependencies and downloads the actual code for each one into the node_modules folder.
What is a linter? (the spell-checker for code)
A linter is a tool that reads your code without running it and flags patterns that are likely bugs or style problems, the same way a spell-checker flags a typo before you hit send.
What is PATH? (why your terminal says command not found)
PATH is an environment variable holding an ordered list of folders your operating system searches every time you type a command name, so a missing folder there is exactly why your terminal says command not found.
What is the terminal, and where do I actually find it?
The terminal is a text-based window where you type commands to your computer one line at a time and it types its replies back, instead of you clicking icons and menus.
Files & project
What all those files in your repo actually are.
What are .cursorrules and .cursorignore? (steering Cursor)
A .cursorrules file used to be Cursor's one file for telling its AI how to work in your project, now replaced by Project Rules (.mdc files in .cursor/rules), while .cursorignore is a separate file that tells Cursor which files to keep out of the AI's context (mostly).
What is .cursorignore? (keeping files out of your AI's sight)
A .cursorignore file at the root of your project lists the files Cursor's AI features are not allowed to read, using the same pattern syntax as .gitignore.
What is .gitignore? (the list of things git pretends not to see)
A .gitignore file is a plain-text list of file and folder patterns that tells git which things to never track, stage, or commit, even when they sit right there in your project folder.
What is a repository? (it's not just a folder)
A repository is your project's folder plus the complete history git has recorded for it, and it can live only on your computer, only on a host like GitHub, or as two synced copies of the same thing.
What is an environment variable? (and why your app has a .env file)
An environment variable is a named value your app reads from its surroundings at runtime, instead of having it hardcoded, so secrets and settings can change without touching a line of code.
What is CLAUDE.md? (your agent's memory file)
CLAUDE.md is a plain markdown file in your project that Claude Code reads automatically at the start of every session, so instructions you'd otherwise repeat every time only need to be written once.
What is package.json? (your project's ID card)
package.json is the file at the root of a JavaScript project that names it, lists every package it depends on, and defines the shortcut commands, like npm run dev, that you and your agent both use to work on it.
Web & deploys
From localhost to the real internet.
Custom domains and DNS (why your site takes hours to go live)
DNS is the address book that tells the internet which server your domain name actually points to, and the wait after you set it up is computers around the world clearing their old cached answer, not your domain "travelling" anywhere.
Localhost vs production: why your app works on your machine but breaks online
Localhost is your app running privately on your own machine; production is the live version everyone else can actually use, and the two don't always behave the same.
npm run dev vs npm run build: why your agent runs one to code and the other to ship
npm run dev starts a local server that reloads instantly as you edit, while npm run build compiles a minified, optimized version of your app for production, and running one is never a substitute for the other.
SSR vs SSG: where your pages get built, and why it matters
SSR, SSG, and ISR are three different moments when your page's HTML actually gets built, on every visit, once at deploy time, or once and then quietly refreshed, and each one trades speed against freshness differently.
What do HTTP status codes like 404 and 500 mean?
An HTTP status code is the three-digit number a server sends back with each response, telling you in one glance whether the request worked, got redirected, or whose fault it was if it failed.
What does deploying actually mean? (from your laptop to the internet)
Deploying is the step where your app stops running only on your own machine and starts running on a server that stays on all the time, at an address anyone can visit.
What is a CDN? (why your site is fast on the other side of the world)
A CDN is a network of servers placed in cities all over the world that hold cached copies of your site's files, so a visitor's browser downloads them from a nearby server instead of the one machine where your app actually lives.
What is a port, and why won't my server start on 3000?
A port is a number from 0 to 65535 that tells a computer which program on it should receive a piece of network traffic, so only one program can listen on a given port at a time.
What is a rate limit? (why the API told you to slow down)
A rate limit is a cap an external service puts on how many requests you can send it per minute, and going over it gets you a 429 error instead of your data.
What is a serverless function, and why did Vercel add api/?
A serverless function is backend code you deploy without managing a server; the platform starts an instance to run it only when a request arrives, then can shut it down, and you're billed for the seconds it actually ran.
What is a webhook, explained simply?
A webhook is a URL you give another service so it can push you a message the instant something happens, instead of you asking over and over.
What is a WebSocket, and why did realtime break on deploy?
A WebSocket is a single connection between a browser and a server that both sides keep open and can send messages over at any time, instead of the browser having to ask for updates one request at a time.
What is an API endpoint? (the doors your app knocks on)
An API endpoint is a specific URL that, combined with an HTTP method, tells a server to do one exact thing, like fetch a user or create a payment.
What is CI/CD? (why your site updates itself after a push)
CI/CD is the automated pipeline that builds and tests your code the moment you push it, then ships it live on its own if everything passes, no one has to click deploy by hand.
What is CORS and why is it blocking your app?
CORS is the browser's rule that blocks your frontend's JavaScript from reading a response from a different origin unless the server explicitly says it's allowed.
What is hosting? (where your app actually lives)
Hosting is the always-on computer, somewhere out on the internet, that keeps your app reachable at a URL even after you've closed your laptop.
What is lazy loading (and why did my images stop appearing)?
Lazy loading defers fetching an image, iframe, or chunk of code until it's actually needed, usually because the user is about to scroll it into view or open the screen that uses it.
Where is the browser console and what does it show?
The browser console is a panel built into every browser's developer tools that prints every error, warning, and logged message your page's JavaScript produces, in plain text you can copy and hand to your AI.
Why do I still see the old version after I deploy?
Browser cache is the storage on your device where your browser keeps copies of a site's files, like HTML, CSS, JavaScript, and images, so it can show the page again without downloading everything from the server, even after the live version has changed.
Databases
Where your data lives, and how to keep it yours.
Lovable Cloud vs your own Supabase: where your data actually lives
Lovable Cloud is a backend Lovable provisions and manages for you on Supabase technology, invisible from your own Supabase account, while connecting your own Supabase project keeps the database, keys, and dashboard under your control.
What is a database migration? (changing the shape of your data, safely)
A database migration is a small, version-controlled file that describes one specific change to your database's structure, applied in order so every environment ends up with the exact same schema.
What is a database schema? (the floor plan of your data)
A database schema is the blueprint of your data: which tables exist, what columns each one has, and how they connect to each other, agreed on before any real row gets stored.
What is a query? (and why does your agent say it failed)
A query is a request or instruction you send to a database, asking it to fetch, filter, or change data, and the same word also names a piece of a URL and a data-fetching library in React apps.
What is an ORM, and why did my AI add Prisma?
An ORM (object-relational mapper) is a library, like Prisma or Drizzle, that translates function calls in your code into SQL, using a schema file as the single source of truth for what your data looks like.
What is RLS (Row Level Security)? The setting that keeps your Supabase data private
Row Level Security is the Postgres rule set that decides, row by row, who can read or write data, and in Supabase it's the only thing standing between your public anon key and your entire database.
What is seeding a database? (fake data, real reasons)
Seed data is a starter set of records, like sample users or products, that a script writes into your empty database automatically, so there's something to look at and test before real users ever show up.
What is Supabase? (the database your AI keeps choosing)
Supabase is a hosted Postgres database bundled with authentication, instant APIs, file storage and realtime updates, which is why most AI app builders wire it up as your backend by default.
Where does your app actually store data? Databases, explained
A database is where your app's information actually lives after you close the tab, organized into tables so it can be found, trusted, and shared by many users at once.
Security
Keys, secrets, and the mistakes that leak them.
API key vs secret vs token: what is the difference?
A secret is any credential you keep hidden; an API key is a long-lived secret identifying your app, and a token is a short-lived secret identifying a user or session.
What is a JWT? (the wristband your app checks at the door)
A JWT is a signed token that proves who a user is, so your server can trust it on every request without looking anything up in a database.
What is HTTPS and why does Chrome say 'Not secure'?
HTTPS is HTTP run inside an encrypted connection (TLS) whose identity is vouched for by a certificate, so your browser can confirm the traffic can't be read in transit and that it's really talking to the site it claims to be.
What is input validation? (and why AI code often skips it)
Input validation is the server-side check that incoming data (form fields, URL parameters, API request bodies) has the type and shape you expect before your app stores it or acts on it.
What is OAuth? (how Sign in with Google actually works)
OAuth is a protocol that lets you grant one app limited access to your account on another service (Google, GitHub, etc.) without ever handing that app your password.
What is prompt injection? (when someone else talks to your AI)
Prompt injection is malicious or misleading text hidden inside a webpage, email, file, or issue that your AI agent reads and treats as an instruction, instead of as content to summarize or check.
What is sandboxing? (the walls around your agent)
Sandboxing runs code in a sealed environment that can't touch your files, network, or accounts unless you explicitly open a door, so mistakes and malicious scripts stay contained.
What is SQL injection? (and why input validation is non-negotiable)
SQL injection is a security flaw where text typed into a form field gets inserted straight into a database query, letting an attacker rewrite what that query actually does.
What is the principle of least privilege? (why your agent keeps asking permission)
The principle of least privilege means every key and account gets the minimum access it needs to do its job, so one leak or one bad command can't take everything down.
What is XSS? Can user input really run scripts in my app?
Cross-site scripting (XSS) is a security vulnerability where an app renders untrusted input as executable code instead of plain text, letting an attacker's JavaScript run in other users' browsers with the same access as the app itself.
What's the difference between authentication and authorization?
Authentication verifies who a user is, usually at login, while authorization checks what that verified user is allowed to do or see, and building only the first one is why any logged-in user can end up reading data that isn't theirs.
Your API key is exposed: what happens, what to do, how to prevent it
An exposed API key is a secret that leaked somewhere public, and from that moment it works for whoever finds it, not just you.
Want these explanations about your own project?
That's what the Glimfly app does: it watches your agent work and explains your sessions live. Join the waitlist to get it first.