MCP App Store
by KeenEthics
Bool icon

Bool MCP Server for ChatGPT

By CodeHS

Overview

Bool turns a description into a working web app, so you can make the tools you need without writing code: a habit tracker, a team CRM, a sign-up form, a project dashboard. ChatGPT builds the app on Bool, with a database and sign-in when it needs them. When a build is saved, the app appears in the conversation, and you can open the full Bool editor beside the chat to see the live preview, its data, and its version history. Ask for changes in plain language, add an image you share or one ChatGPT generates, set the app's icon, and publish it to a live URL when it's ready. Bool works on the projects in your connected Bool account. Changes are saved as versions you can restore, and nothing goes live until you ask to publish. Bool doesn't process payments in ChatGPT.

How to connect

  1. ChatGPT: Open the Bool listing in the ChatGPT app directory and choose Connect. ChatGPT listing Bool exposes 37 tools here.

Tools

add_image

ChatGPT
Save an image into a Bool project's files, such as one the user shared in the conversation or one you generated. Files under public/ are served from the app's root, so public/hero.png is /hero.png. Replaces a file already at that path. For the app's icon, use set_project_icon.

build_app

ChatGPT
Build a Bool project and return the compiler output. ok is false when the build fails. The output also lists runtime errors the project's open preview reported since the last build.

cancel_project_turn

ChatGPT
Stop the AI turn currently building or editing a project. This can leave partial edits from the interrupted turn, so use it only when the user asks to stop or cancel the active build.

create_file

ChatGPT
Create a file in a Bool project, or replace an existing file's whole content.

create_project

ChatGPT
Create a blank Bool project from a starter template, to build yourself with get_build_guide and the file tools. It is created in the workspace chosen in Bool's settings unless workspace_id names another (see list_workspaces).

create_records

ChatGPT
Insert one or more rows into a project's entity. Pass each row's declared fields — id and created_at are filled by the database. Returns the created rows. Writes as the project admin. PRIVATE entities: owner_id (which end-user owns the row) is NOT NULL and normally set from the signed-in user — the admin path has no signed-in user, so you MUST include owner_id in each row (an end-user's id) or the insert fails. Public entities need no owner_id. Max 50 rows per call.

define_entity

ChatGPT
Declare or update a project's data model. This replaces the entity schema and generated TypeScript declaration files, then applies an additive database migration: it can create a table and add missing columns, but never drops columns or changes their database types. Choose access when creating the entity: 'private' (default) gives rows per-user isolation, while 'public' creates one shared table; changing an existing private entity to public is not supported. Managed columns are added automatically and MUST NOT be declared: id (uuid), created_at, and — for private entities — owner_id. After changing models, re-run bool types in a linked repo to refresh local types.

delete_file

ChatGPT
Delete a file from a Bool project.

delete_project

ChatGPT
Permanently delete a project, including its database, files, deployments, screenshots, and attachments. If deployed, its live app is removed. Owner-only. There is no undo.

delete_record

ChatGPT
Delete one row of a project's entity by id. Deletion is permanent — there is no undo.

edit_file

ChatGPT
Edit a file in a Bool project with find-and-replace. An edit fails unless its old_str matches exactly one place in the file, so include a few lines of surrounding context. Pass several edits in one call to apply them in order.

fork_project

ChatGPT
Remix an existing project: create a copy of it in your workspace. By default the project's structure (entities, files, and code) is copied; set with_data to true to also copy all database rows, including private end-user rows. with_data requires owner access.

get_build_guide

ChatGPT
Return the build guide for a Bool project: the instructions Bool's own builder follows for that project. It covers the Bool SDK APIs the project has, how data, sign-in and files work, which files are locked, the design rules, and how the file, build and version tools fit together.

get_editor_session

ChatGPT
Return the session the Bool editor panel uses to load the editor. Called by the panel, not the model.

get_project

ChatGPT
Fetch a project's live URL and its deployment history (every version, which one is currently live). The public URL is stable across versions. Use update_project to change visibility or other project details.

get_project_connection

ChatGPT
The public connection descriptor a local bool-sdk client needs to use this project as its backend: slug, appOrigin, the project's schema, and the shared Supabase URL plus anon key. The anon key has no data grants; the gateway holds the real credential. Pair with bool link or hand-write bool.config.json from this. Create admin data keys in Bool's project settings rather than exposing credentials through an AI conversation.

get_project_status

ChatGPT
Check whether an AI turn is currently running for a project and get its current live URL, visibility, and deployment state. Use this to poll after prompt_project until the turn finishes and the project is deployed.

list_entities

ChatGPT
List a project's declared entities (data models): each entity's name, access mode ("private" = per-user rows via row-level security, "public" = one shared table), and fields (name, type, required, enum values). Entities are how a Bool project stores data — apps read/write them via bool.entities.<name> (SDK) or the record tools here. Use define_entity to create or extend one.

list_files

ChatGPT
List every file in a Bool project with its size.

list_projects

ChatGPT
List the Bool projects across all your workspaces (personal and shared). Returns each project's id, name, slug, workspace, whether it's deployed, and its visibility (liveAccess — who can view the deployed app: "link" public, "workspace", or "restricted"; and editorAccess — who can open it in the editor). Pass workspace_id (from list_workspaces) to limit to one workspace. (A workspace-bound access token always lists only the workspace it's bound to.)

list_records

ChatGPT
Read rows from one of a project's entities (or any table in its database). Returns the matching rows plus the total count. Supports column filters (eq, neq, contains, starts_with, gt/gte/lt/lte, is_null/is_not_null — all AND-ed), a sort, and limit/offset paging (limit caps at 500). This reads as the project admin: it sees ALL rows, including every end-user's rows on a private entity.

list_templates

ChatGPT
List the starter templates you can pass to create_project's template parameter. Defaults to vite-react.

list_workspaces

ChatGPT
List the workspaces you can act in, including your original workspace and any additional workspaces you belong to. Each entry includes plan (the workspace's plan tier) and paid (true for a paid plan, which lifts the free deployed-bool cap). Use a workspace's id as the workspace_id argument to other tools (create_project, list_projects, publish_project, fork_project, move_project) to target it. A workspace-bound access token sees only the workspace it is bound to.

move_project

ChatGPT
Move an existing project to a different workspace. Only the project owner can move it. The project keeps its current visibility and editor settings; if the current visibility is not allowed in the new workspace, use update_project to change visibility first. The project's folder is cleared so it lands at the root of the new workspace. Use this to fix workspace placement, e.g. when a work project was created in a personal workspace.

open_bool_editor

ChatGPT
Show the Bool editor for a project: its live preview, files, data and versions. The preview follows edits made with the Bool file tools. Without project_id it shows the user's projects.

prompt_project

ChatGPT
Send a prompt to an existing project's AI and run a headless turn in the background. The AI reads the project, edits files, runs build_app, and — unless auto_publish is false — publishes the result when the turn completes. Before prompting, check the project's current workspace. If the prompt is clearly for the wrong workspace (e.g. a work request on a project that lives in a personal workspace), move it to the right workspace with move_project first. Use this to iteratively build and refine a website or app. Poll get_project_status to wait for the turn to finish.

publish_project

ChatGPT
Build and publish a project (or a specific snapshot) so it goes live at its deployed URL, governed by the project's current access setting. If no snapshot_id is given, the project's current source is published, including any edits made with create_file or edit_file. A clear error is returned if the workspace's deployed-bool cap is reached. Before publishing, consider whether the project is in the right workspace for its content. If it looks like a work project in a personal workspace, use move_project to relocate it to a shared/company workspace first.

read_file

ChatGPT
Read a text file from a Bool project. Long files come back in pages; pass nextOffset to read the next page. Binary files return their size instead of their bytes.

render_project_widget

ChatGPT
Render an interactive status card for one Bool project: build status, the live app and editor links, and follow-up controls. It's for clients that can't show the Bool editor; in ChatGPT, save_version and open_bool_editor already show the project. The card refreshes itself, so one call is enough.

run_command

ChatGPT
Run a command, such as npm install, in the Bool project's own sandbox and return its exit code and output. The dev server is already running, and background commands are not supported.

run_db_migration

ChatGPT
Run SQL in the Bool project's own database schema, inside a transaction, and return any result rows. On failure the result includes Postgres's error detail and hint. To create tables, define_entity is usually the better choice.

save_version

ChatGPT
Save the project's current source as a version in its version history, after build_app passes. The user sees it in the editor's version list and can restore it later. Does nothing if the source hasn't changed since the last version. Does not publish; use publish_project for that.

set_project_icon

ChatGPT
Make an image the app's icon, the one browsers and home screens show, replacing the current icon. Bool writes every icon size from it. Use a square image, such as one the user shared or one you generated.

settings_read

ChatGPT
Return Bool's plugin settings, their current values, and their layout.

settings_update

ChatGPT
Change Bool's plugin settings.

update_project

ChatGPT
Rename a project, set its description, or change its visibility. Setting visibility to link can make a deployed app publicly reachable. Owner-only. Visibility changes are gated by the workspace's plan, the same way as the dashboard.

update_record

ChatGPT
Patch one row of a project's entity by id: sets the given fields, leaves the rest untouched. Returns the updated row.

Capabilities

WritesInteractive

Example Prompts

Click any prompt to copy it.

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/bool"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/bool)