Every line-of-business application is the same skeleton wearing different data: tables, screens to edit them, an API, users and roles, a log of who changed what. GoatCheese makes that skeleton a build product. You write one schema that names your entities, their fields and who may touch them; one command turns it into a working, documented application. Your effort goes into the ten percent that is actually your business.
The schema is the single source of truth. Everything downstream is derived from it and can be derived again.
Illustrative session — output condensed. GoatCheese is a command-line tool; everything it does is a command you can script, run in CI, or hand to an AI assistant.
One readable file: the entities (clients, quotes, projects…), their fields and types, how they relate, and which roles may read, write or delete each one. It is the model of the business, not a pile of configuration — a product owner can follow it.
gc build. The engine reads the schema and produces the whole application: database, admin screens, API, authentication, permissions, audit trail, an MCP server for AI assistants, deployment and backup commands, and the documentation to go with it. Minutes, not sprints.
Business rules — pricing, approvals, notifications, connectors to other systems — live in hooks and wrapper classes that sit beside the generated code, never inside it. Change the schema, rebuild, and your logic is still there.
Each piece below is generated from the schema, consistent with every other piece, and thrown away and rebuilt whenever the model changes.
Tables, keys, relations and indexes created from the schema; migrations when it changes. No hand-written table definitions to drift out of sync.
Lists, forms, filters, search and exports for every entity, shaped by the same roles that guard the API. Ready on day one, refined later.
A consistent endpoint per entity — list, read, create, update, delete — each route checked against the role matrix before it runs.
Sign-in with short-lived tokens and rotating refresh, role-based access down to entity and field, and optional OAuth 2.1 so other applications and AI assistants log in as a named user.
Who did what, to which record, when — recorded on every write, in the admin and through the API alike. The question "who changed this?" has an answer.
The application is exposed to AI assistants through the Model Context Protocol: the same entities, the same permissions, reads separated from writes, writes confirmed before they execute.
Every build and deploy ends with an audit: access-rule coverage, headers, cookie flags, secrets, file permissions, dependencies. Findings land in a report before they land in production.
Commands to ship a release, run migrations and take or restore a backup — the operational chores are part of the build, not an afterthought.
Reference docs for the data model and the API, regenerated with every build, so what is written down is what actually runs.
Code generators have a bad reputation: fast on day one, unmaintainable by month six. These constraints are what make the difference.
Every APIgoat product and most client back offices of the last few years come out of the same engine. Each one exposes its own MCP server.
Multi-tenant AI chatbot platform: clients, document libraries, keys, usage billing and a client portal — all schema entities, plus retrieval logic in hooks.
product page →Invoicing and time tracking for freelancers: clients, projects, time logs, expenses with receipt scanning, clean invoices.
billboy.apigoat.com →Contacts, companies, quotes, invoices, payments and document storage for a field-service company — steered day to day from an AI assistant.
Paper-trading crypto bot platform: runs, strategies, decisions and P&L as entities; the daemons and market feeds as custom services beside them.
Accounting and time tracking: clients, suppliers, bills, payments, expenses, projects — the engine running its author's own books.
A voice-first French tutor for kids with a mobile app and a parent admin — the back office and progress data generated, the voice sessions custom.
Eighteen projects on the engine so far: dashboards, client portals, a family organiser, a raid planner, a subscription manager, a lottery back office. Whatever the schema describes.
A custom CRM, ERP module, portal or back office on GoatCheese starts with a working, documented, AI-ready application in the first week — and the budget goes to what makes it yours. Send a brief; get a shape and a quote within one business day.