You do not need to memorize developer vocabulary to customize your marketplace with AI. However, understanding a few recurring terms makes it easier to follow what the AI is doing, notice when it has misunderstood you, and make decisions before a change is published.

This glossary complements our Technical terms and definitions, which explains Sharetribe concepts such as the frontend, backend, APIs, environments, listings, and transaction processes. Here, we focus on the general words that appear in AI coding tools, and on the places where those words meet Sharetribe specifics: which credentials are secret, where your marketplace texts live, and which half of the template can be trusted with sensitive logic.

This glossary also complements our tutorial on how to start AI-assisted coding on top of Sharetribe, our what you can and can’t do with vibecoding article, and how to succeed with AI-assisted coding (vibecoding) on Sharetribe.

Note that you do not need to read this glossary end to end to get use of it. Instead, it might be better to keep it open in the background (or return to it) in your browser, and search for a specific term whenever you encounter one you don’t recognize.

You can also ask the AI: “Explain <insert term> in plain language, tell me why it matters for this task, and tell me what you want me to do.”

When one of these words appears in a request from the AI, read the request twice and ask what will change before you approve it:

reset, force, delete, remove, overwrite, discard, secret, token, key, .env, Live, production, payment, Stripe, refund, authentication, login, transaction process, privileged, Integration API, database, migration, upgrade.

AI and coding assistants: Agent, Model, Prompt, Context window, Tokens, Plan mode, Permission mode, Plugin and skill, Slash command, MCP server, Session, Instruction files, Hallucination.

Your computer and local preview: Terminal, Project folder, Node.js, Yarn, Dependency, node_modules and package.json, localhost:3000, Development server, Browser console, WSL.

Git and GitHub: Git, Repository, Clone, Origin and main, Commit, Diff, Branch, Merge, Merge conflict, Pull request, Push and pull, Upstream, .gitignore, GitHub authentication, Revert and roll back, Destructive commands.

Configuration, release, and quality checks: Environment (three meanings), Credentials, .env file, Environment variable, Sandbox and test mode, Build, Deploy, render.yaml, Staging, Production or Live, Server, Server-side rendering, Content Security Policy, HTTP status codes and CORS, Test, Error message and stack trace.

Words the AI uses when describing its changes: Component and route, Refactor, Hardcode, Regression, Edge case and happy path, Translation files, Database and migration.

What the AI means: An AI that can inspect files, use tools, and carry out several steps toward a goal, rather than only answering a question.

What you should do: Give it a specific outcome and boundaries. Review changes and test the result; an agent can act, but it can still be wrong.

What the AI means: The underlying AI system producing the answer or doing the work. Different models may vary in speed, cost, and ability. Examples include Fable, Opus, and Sonnet (from Anthropic’s Claude 5) and Astra, Sol, Terra, and Luna (from OpenAI’s ChatGPT 5 and 6).

What you should do: Use a capable model for complicated changes, debugging, or reviews. A smaller model may be enough for a simple visual adjustment.

What the AI means: The instruction or request you give the AI, including any examples, constraints, and background.

What you should do: Describe the result you want, who it is for, where it should appear, and how you will know it works.

What the AI means: The amount of conversation, code, and other information the model can consider at one time. When a conversation grows long, some tools automatically “compact” or summarize it, and details can be lost in the process.

What you should do: Keep each task focused. When a session becomes long or the tool reports that it compacted the conversation, restate the rules that matter, or start a fresh session with concise project notes and the relevant files.

What the AI means: Small units used to measure the text an AI reads and produces. Usage limits, credits, rate limits, and charges are usually based on tokens.

What you should do: Do not try to count them yourself. Keep requests focused and avoid repeatedly loading large files that are not relevant, such as the marketplace text files.

What the AI means: A mode in which the AI investigates and proposes an approach before editing files.

What you should do: Use it for an unfamiliar or multi-step change. Read the plan and correct misunderstandings before allowing implementation.

What the AI means: A setting that controls which actions the AI can take automatically and which require your approval.

What you should do: Start with a mode that asks before sensitive actions. Read requests carefully, especially before publishing, overwriting files, or using credentials.

What the AI means: A plugin is an installed package that can add tools or instructions to an AI tool. A skill is a set of instructions for a particular task inside a plugin or AI tool. The Sharetribe skills plugin contains skills for setting up the Sharetribe Web Template, connecting it to GitHub, deploying it to Render, and building features.

What you should do: Use the Sharetribe skills plugin for its guided workflows, starting with /start. Let the AI choose the relevant skill, and still review the proposed result.

What the AI means: A short instruction that starts with a forward slash, such as /start, which triggers a specific workflow or setting inside the AI tool instead of being read as a normal message.

What you should do: Use /start from the Sharetribe skills plugin to begin setup, connect GitHub, deploy to staging, or build a feature. If a slash command is not recognized, the plugin may need reloading (/reload-skills in a new conversation) or reinstalling.

What the AI means: A small program or online service that an AI tool connects to, giving it access to an outside source of information or extra abilities. What it can do depends on that server.

What you should do: Connect your coding tool to Sharetribe’s documentation MCP server so it can search current technical documentation while it works. Check what any other MCP server can access before enabling it.

What the AI means: One continuous conversation with the AI, including everything it has read and done so far. A fresh session starts with no memory of earlier work. A handoff prompt is something you can ask for when switching to a fresh session, if you want to maintain some context from the old one.

What you should do: Start a fresh session when the AI begins mixing up earlier instructions with the current task. First save your work and ask for a short handoff summary: what was done, what was decided, what is still open, and the next step.

What the AI means: Files in the project that the AI reads automatically at the start of a session. The Sharetribe Web Template ships an AGENTS.md file that describes its structure and conventions. Some tools use their own names, such as CLAUDE.md or a rules folder.

What you should do: Add a short section about your own marketplace: what must not be touched, which environment is which, your staging address, and your testing rules. It saves you from repeating the same instructions in every session.

What the AI means: An answer that sounds certain but is invented or wrong: a setting that does not exist, a command with the wrong name, or “fixed” when nothing changed.

What you should do: AI can sometimes hallucinate. Audit the changes you’ve made and features you’ve implemented in a separate conversation with the AI, either individually or in batches, to avoid problems down the line. Giving the AI the Sharetribe documentation leaves it fewer reasons to guess. Further, consider adversarial iteration to combat bugs and hallucinations in your AI’s implementation.

What the AI means: The terminal is a text-based window where you or the AI can run instructions on your computer. A command is one such instruction, such as starting the marketplace preview, or set up a development environment.

What you should do: You do not need to memorize commands. Ask the AI to explain what a command will do before running it if the effect is unclear. Run commands in the folder the instructions specify, and if one fails, share the complete error message with the AI instead of guessing at a fix.

What the AI means: The project folder (or working directory) is the folder the terminal and coding tool are currently working in. A path is the written address of a file or folder on your computer; short paths are read from the project folder.

What you should do: Confirm that the coding tool has the correct marketplace folder open before installing, editing, or running anything. When the AI mentions a path, ask it to open the file or point you to it, and avoid moving project files outside the expected folders.

What the AI means: Software that runs JavaScript, a programming language, outside the browser. The Sharetribe Web Template uses it both for development tasks and for its own small server. You may also see it called the runtime, meaning the software environment the application runs in.

What you should do: Let the setup workflow install or check the required version. Do not remove or upgrade it mid-project merely because the AI mentions a newer release, and keep the version your hosting provider is set to use in line with the one on your computer.

What the AI means: A tool that downloads the project’s packages and runs named tasks. Yarn is the package manager used in Sharetribe’s setup instructions.

What you should do: Let the AI run the project’s documented Yarn commands. Avoid switching package managers, because doing so can create conflicting dependency files.

What the AI means: A reusable package of code that the project relies on. A dependency can be “deprecated” (no longer maintained) or have a known “vulnerability” (a security weakness); the template’s audit check lists packages with known problems.

What you should do: Ask why a new dependency is needed before adding it. Prefer capabilities already present in the Web Template when they solve the problem, and give the AI the audit results when it reports a vulnerability.

What the AI means: A web address that points to a service running only on your computer. 3000 is the port, which identifies the particular local service. The error “port 3000 is already in use” means a development server is already running somewhere.

What you should do: Open it while the development server is running. If it does not load, ask the AI to check the server and the address shown in the terminal. If the port is already in use, ask the AI to find the running server rather than starting a second one.

What the AI means: A program that runs your marketplace locally and refreshes it while you work. The automatic refresh after a file changes is called hot reload. In the Sharetribe Web Template, the command yarn run dev starts the fast local preview, and yarn run dev-server runs the real server so you can check server-side rendering before deploying. Read more about the development environment setup here.

What you should do: Keep its terminal session open while reviewing changes. Wait briefly after a change; if the preview does not update, refresh the page or ask whether the server needs a restart. A local development server is for building and testing, not for serving your live marketplace.

What the AI means: A panel in your web browser (open it with F12, or right-click the page and choose Inspect) that shows errors from the page, the requests it makes, and its structure.

What you should do: When a page misbehaves, open the console, copy the red error messages, and give them to the AI together with the page address. Blocked scripts, missing images, and failed requests usually show up here first.

What the AI means: A way to run a Linux environment inside Windows. The Sharetribe Web Template’s own instructions recommend it for Windows users.

What you should do: If you are on Windows, tell the AI at the start of every session whether you use WSL or plain Windows, so the commands it suggests match your setup.

What the AI means: A system that records changes to the project over time, making it possible to compare versions and recover earlier work.

What you should do: Have the AI save a clear checkpoint after each small change that you have tested.

What the AI means: The project’s files together with their Git history. Git is the tool on your computer; GitHub is the website where a remote copy of the repository lives, acting as an online backup and a collaboration point.

What you should do: Keep your marketplace repository private unless you intentionally want to publish the code. Never include secret .env files. Your GitHub repository is your backup, so push regularly.

What the AI means: To download a repository and its history to a new folder on your computer.

What you should do: Clone once when setting up the project on a computer. Open the newly created project folder before continuing.

What the AI means: Origin is the name Git gives your own repository on GitHub. Main is the primary branch of your project. A command such as git push origin main sends your main branch to your GitHub repository.

What you should do: Expect these two words in almost every Git command. In the Sharetribe setup, origin is your repository and upstream is Sharetribe’s template repository.

What the AI means: A commit is a saved checkpoint in the project’s Git history, with a short message describing it. Uncommitted changes are edits on your computer that have not yet been saved as a checkpoint.

What you should do: Commit after a change works. Use a message describing the user-facing result, such as “Add provider details to listing cards.” Committing allows you to return to this checkpoint if anything goes wrong in a future update. Before changing direction, ask the AI to show what is uncommitted and either save or discard it deliberately.

What the AI means: A view showing exactly what was added, removed, or changed between two versions.

What you should do: Review the diff after a focused change. Ask the AI to explain important changes in plain language and remove unrelated edits.

What the AI means: A separate line of work used to try a change without immediately altering the main version.

What you should do: Use a branch for a larger or uncertain experiment. Ask which branch is active before the AI begins.

What the AI means: To combine changes from one branch or version into another.

What you should do: Merge only after reviewing and testing the change. The AI can perform the operation, but you should understand which versions are being combined.

What the AI means: A situation where two versions changed the same part of a file and Git cannot automatically decide which one to keep.

What you should do: Do not choose blindly. Ask the AI to explain both versions and confirm that the resolved result preserves your custom behavior.

What the AI means: A page on GitHub for proposing, reviewing, and discussing changes before they are merged.

What you should do: You only need one when another person reviews your work, such as a developer or a Sharetribe Expert. Working alone with the Sharetribe skills plugin, your changes go straight to the main branch.

What the AI means: Push sends code changes from your computer to a remote repository, such as after you’ve completed a feature. Pull brings remote commits to your computer, such as when reverting to an earlier version or applying changes made by another developer into your local environment.

What you should do: Commit local work before pulling changes from the remote repository.

What the AI means: The original repository from which your project may receive updates, such as the Sharetribe Web Template repository. Its new versions are called releases, the list of what changed is the changelog, and a change that can break customized copies is a breaking change.

What you should do: Treat upstream updates as proposed changes. Sharetribe publishes new template versions often, typically several times a month, so pick the releases that matter to you, and test any merge with your customizations.

What the AI means: A list of files and folders Git must never track, such as the .env file (since it often contains important “secrets”, i.e., personal tokens for your accounts) and the node_modules folder.

What you should do: Before the first push to GitHub, ask the AI to confirm that .env is listed in .gitignore and that no secrets are in the commit. Anything that has ever been pushed should be treated as public.

What the AI means: Proof to GitHub that you are allowed to push to your repository. It can be a browser login, a personal access token, or an SSH key.

What you should do: When the AI says it cannot push and mentions authentication, complete the login steps yourself rather than pasting an application token into the chat. These tokens are secrets; treat them like the client secret. These are not to be confused with the tokens that the AI consumes, as mentioned earlier in this guide.

What the AI means: Revert means undoing a particular code change by adding a new commit that reverses it, so the history stays intact. Roll back usually means returning to an earlier state more broadly, for example redeploying an earlier build at your hosting provider without changing code.

What you should do: Identify the exact working commit first. When a deploy has broken staging or Live, rolling back to the previous deploy at the hosting provider is often faster than fixing code. Reverting code does not automatically undo data or Console changes made at the same time.

What the AI means: Commands that throw away work or rewrite history. git reset --hard discards uncommitted changes, git checkout . discards edits to files, git push --force overwrites the history on GitHub, and deleting a branch removes work that was never merged.

What you should do: Ask the AI to explain exactly what will be lost before you agree. Prefer a commit first and an undo second, so the discarded state can still be recovered.

What the AI means: In Sharetribe, an environment is a marketplace instance: Dev, Test, or Live. In the coding tool, the environment setting says where the AI works, on your computer or in a cloud sandbox. An environment variable is a configuration value such as a client ID.

What you should do: When the word appears, ask which one is meant. Your local setup and your staging app use Dev environment credentials. Live credentials are used only in the production deployment.

What the AI means: The values that identify your marketplace to Sharetribe, Stripe, and other services. Some are public by design: the Sharetribe client ID and the Stripe publishable key. Others are secrets: the Sharetribe client secret, the Stripe secret key, Integration API credentials, and any access token.

What you should do: Type secrets only into the terminal prompt that asks for them, the .env file, the hosting provider’s environment variables, or Console. Never paste them into a chat message or a screenshot. Copy them from the right Sharetribe environment (or a corresponding third party website): On Sharetribe, dev for local work and staging. If a secret leaks, create a new application in Console and delete the old one.

What the AI means: A usually hidden file that stores local configuration values, including credentials that should not be committed to Git. The setup workflow creates it for you.

What you should do: Let the setup workflow create it. Never paste its contents into an AI chat or publish it to GitHub. On Windows the file is visible; on Mac and Linux, files whose names start with a dot are hidden by default, so ask the AI how you can open it on your machine if you need to see or edit it.

What the AI means: A named setting supplied outside the code, such as a client ID, secret, or service address. A .env file is one way to provide these locally; hosting providers have their own settings screen for the same values.

What you should do: Enter each value in the location specified for Dev, staging, or Live. Ask before copying a value between environments. Changing a variable at the hosting provider requires a redeploy before it takes effect.

What the AI means: Stripe’s test environment, where payments use test card numbers and no real money moves. It is separate from Sharetribe’s Dev and Test environments, although they are used together.

What you should do: Use Stripe sandbox keys with your Dev environment and Stripe test cards for every payment test. Live Stripe keys belong only in Console’s Live environment.

What the AI means: A process that turns the source code into the files a server can run. “The build failed” means the preparation stopped because of an error. A build catches broken code and missing files; it does not prove that a feature behaves correctly. Some projects run automatic checks on every push, called continuous integration (CI); the Sharetribe setup does not require this.

What you should do: Give the AI the complete build error. Fix the underlying problem and run the build again before deploying.

What the AI means: To publish a particular version of the code to a hosting service so it can run online. A redeploy publishes again, for example after changing an environment variable. Learn more about deploying here.

What you should do: Confirm the target is staging or Live and which commit is being deployed. Test staging before releasing to real users.

What the AI means: A file the Sharetribe skills plugin creates in your project that describes your staging service on Render: its name, build command, and the environment variables it needs.

What you should do: Keep it committed. Secrets are marked to be entered in the Render dashboard, not stored in the file. If the service name is not unique, Render adds random characters to the address, and the file has to be updated to match.

What the AI means: An online version used to review code before it becomes the live marketplace. In the Sharetribe setup, the staging app on Render is connected to your Dev environment with the Dev client ID and secret.

What you should do: Restrict access with the basic authentication username and password variables while the marketplace is under construction. Do not treat successful staging checks as permission to skip final Live checks.

What the AI means: The customer-facing version of the application, where mistakes affect real users and transactions. It is connected to your Live environment, which must be switched to custom code mode in Hosting mode before it can use your own code.

What you should do: Publish only a tested, committed version. Be especially careful with payments, authentication, private data, and transaction behavior.

What the AI means: A program that responds to requests from browsers or other software. The Sharetribe Web Template has its own small server, separate from Sharetribe’s backend. It is the only trusted place in your code for secrets, price calculations, and Integration API calls, because anything that runs in the browser can be seen and changed by users.

What you should do: When the AI proposes handling prices, secrets, or Integration API calls in the browser, ask it to move that logic to the server. Choose hosting that can run the template’s Node.js server; not every website host can.

What the AI means: Creating the first version of a page on the server before sending it to the browser. The Web Template uses this for performance and search visibility. An error mentioning “window is not defined” usually means new code assumed it was running in a browser.

What you should do: Preserve the existing setup unless you have a strong reason to change it. Run yarn run dev-server before deploying to catch server-side rendering errors early.

What the AI means: Rules in the template’s server that tell browsers which outside scripts, images, and connections a page may load. Anything not on the list is blocked.

What you should do: If a chat widget, analytics script, or font works on your computer but not on staging, expect a CSP error in the browser console. Ask the AI to add the exact domain to the policy in server/csp.js rather than switching the policy off. Report mode shows violations without blocking; block mode is the setting for Live.

What the AI means: Short numeric results of a request. For example, 404 means not found. You may encounter these errors especially when dealing with live, online environments.

What you should do: Give the code, the page address, and the console message to the AI, and ask what it is about. You can also ask Sharetribe support, if needed. A CORS error is often a sign that the call belongs in the template’s server, not in the browser.

What the AI means: A manual or automated check that expected behavior still works. Lint is an automated check for suspicious code and inconsistencies. Most of the template’s automated tests are snapshot tests, which compare how a component looks with a saved reference; “updating the snapshot” accepts the new look and is not a backup.

What you should do: Test the changed feature and the related customer and provider flows yourself. Ask the AI to run the automated checks after code changes and address relevant findings. Passing checks are useful evidence, but they are not a complete review.

What the AI means: Information recorded when something fails. A stack trace lists the sequence of code involved in the failure. Errors appear in three places: the terminal, the browser console, and your hosting provider’s logs.

What you should do: Share the full relevant message with the AI, along with what you did just before it appeared and where you saw it. Remove passwords, keys, and private user information first.

What the AI means: A component is a reusable building block of a page, such as a listing card or the search bar. A route is a page address pattern, such as /s for search or /l/ for listing pages.

What you should do: When the AI names a component, ask which page and which part of the screen it means. New pages need a route; unless you specifically want to change them, consider asking the AI to leave existing routes unchanged so links and search engine results keep working.

What the AI means: Restructuring code without changing what it does for users, usually to make it tidier or easier to maintain.

What you should do: Ask the AI not to refactor unrelated code as part of your change, unless you have a specific reason to do it. Refactoring makes the diff longer and harder to review, may make it harder to debug and it can introduce regressions without adding anything you can see.

What the AI means: To write a fixed value directly into the code, such as a color, a text, or a listing type name, instead of reading it dynamically from configuration.

What you should do: Ask the AI to use the values that come from Console (hosted configurations) whenever they exist. A hardcoded value stops responding to Console changes and has to be found and edited in code later, so in most cases, it should be avoided - unless there is a specific reason to do it.

What the AI means: A feature that used to work and stopped working because of a later change.

What you should do: After each change, re-test the flows near it, not only the feature you asked for. If you find a regression, tell the AI which commit last worked so it can compare the two versions.

What the AI means: The happy path is the flow where everything goes as expected. An edge case is an unusual situation: an empty field, a very long title, a double click on the pay button, a booking that ends at midnight.

What you should do: Ask the AI to list the edge cases before it builds, and test the important ones yourself. AI-written code tends to cover the happy path first.

What the AI means: JSON files in the src/translations folder (en.json and others) that hold the template’s default texts. In a Sharetribe marketplace, the texts users see come from Marketplace Texts in Console, which override these files.

What you should do: Change texts in Console, not in these files. Editing them through an AI uses a lot of credits on very large files, and the change may be overridden anyway. The one expected exception is when the AI adds a brand-new text for a new feature.

What the AI means: Where an application stores its data and how that data is structured. Vibecoding tools such as Lovable or Bolt add a database (for example Supabase) to nearly every project.

What you should do: Your marketplace data already lives in Sharetribe and is shaped by listing fields, user fields, and extended data, so there is no database for you to set up or migrate. Treat a suggestion to add one as a sign that the AI has misunderstood the architecture, and point it to the Sharetribe documentation.

Technical words can mean slightly different things in different tools. Ask the AI to point to the exact file, screen, or action it means. A good explanation should tell you what will change and what decision, if any, you need to make.

If you're still mostly starting out, you may find the article on what you should know about custom coding as a no-code founder useful.

For the setup itself, follow How to start AI-assisted coding on top of Sharetribe for non-developers. If you're not sure yet what you can achieve with vibecoding, read this article on what you can and can't do.

For broader advice on working with an AI coding tool, read How to succeed with AI-assisted coding on Sharetribe. For Sharetribe-specific concepts, return to Technical terms and definitions.