The Graph page can now narrow a large project down to what you are looking at. Search for models by name, description or field; pick a group of models that share a name (such as every Cashflow model); or focus on one model and the models one, two or three links away. Shift-click a model in the graph to focus it, or open the graph from a model’s page with its new Graph link.
Matches and the focused model are highlighted, linked models can be shown alongside a search, fields can be hidden to see just the structure, enums can be drawn with their values, and the graph now fits the page width.
Two more formats to import: Prisma schemas and Protocol Buffers.
Prisma import reads models, types and enums with their @map names, ids, relations (the foreign key becomes an id that refers to its model), unique fields, defaults and doc comments. Protobuf import reads messages and enums, nested ones included, with optional, repeated, map, oneof, json_name, number sizes, well-known types and comments.
Both generators were reworked so their output is valid and imports back: Prisma now writes each model once with a single @id, valid names mapped to the originals, relations with the back-relation Prisma needs, and correct number types; Protobuf keeps names it cannot spell in json_name and writes sized number types. Also fixed: Microsoft Graph crashed on title and URL fields and named every property after its model, Python files now end in .py, and TypeORM is listed as a database schema.
The new Checks page looks over a project for mistakes: types, references and map keys that point at nothing, duplicate fields and enum values, defaults that do not fit their field, unused or empty enums, fields that no language generates, conflicting number sizes, field names that break their model’s naming style, and missing descriptions. Each finding links to the model, enum or constant to fix.
AI assistants connected over MCP can run the same checks with the new lint_project tool.
New language for code generation and import: Rust, as serde structs and enums.
Field and value names that Rust spells differently keep their original names in JSON through #[serde(rename)]. Optional fields become Option, arrays Vec and maps HashMap; number sizes, dates (chrono) and defaults carry over, and a model that contains itself is boxed. Importing reads serde structs and enums back, including rename and rename_all.
A connected repository can now receive pushes as a pull request instead of commits straight to its branch. Turn on Push as a pull request in the repository settings.
Each push puts every generated file in one commit on a unifiedmodels/<project> branch and opens a pull request that lists what changed, with breaking changes first. While the pull request is open, later pushes update it; once it is merged, the next push starts a new one. With auto-push on as well, every save keeps the pull request up to date and nothing reaches your branch until it is merged.
Every change to a project is now recorded. The new History page lists its versions, shows what each one changed or everything that changed since, with breaking changes such as removed fields, changed types and removed enum values called out, and restores any version in one click.
AI assistants connected over MCP can use the same history with the new list_versions, diff_version and restore_version tools.
Importing into a project now previews a merge: each model, field and enum is marked New, Changed or Unchanged against what the project already has, with the exact changes listed. Applying adds and updates only what you select, keeps everything the import does not mention, and no longer creates duplicates.
Defaults now reach TypeScript, SQL, C#, PHP, Ruby, Go and Swagger, and the URL and Email modifiers reach OpenAPI, Swagger, JSON Schema and Typebox.
JSON Schema output is now a real draft 2020-12 schema, and Go output compiles: a package named after the folder it is pushed to, exported fields with json tags, and enum values kept as entered. Fixed output that did not compile in Typebox, SQL, C# and PHP, and importing JSON Schema and Go keeps names, enums and formats.
UnifiedModels now has an MCP server. Connect Claude, ChatGPT, Cursor, VS Code, Codex and other AI clients to read and edit your models, enums and constants, generate code in 30 languages, import schemas and push to GitHub. Sign in with your UnifiedModels account; no API key needed.
Find setup steps for each client on the MCP page, or from Connect AI (MCP) in the profile menu inside the app. We also published our Privacy Policy.
You can now import from Go, Ruby, BigQuery, F#, AsyncAPI, DynamoDB, Google Discovery, GraphQL, JSON Schema, OpenAPI, RAML, PHP, and Swagger
New languages for code generation: RAML, Ruby, BigQuery, F#, AsyncAPI, DynamoDB, PHP, Mermaid, and Google Discovery.
New paths for easier navigation in the app.
New language for code generation: TypeORM
New language for code generation: Prisma
New language for code generation: Microsoft Graph Schema
New language for code generation: Zod
New language for code generation: Typebox
Introducing the concept of 'Constants'. Data that you can create based on a Model. Currently only available for Typescript and Golang.
Automatically push updates to a GitHub repository when the model changes.
New language for code generation: Swagger
Field types:
Field Options:
New language for code generation: OpenAPI
Field types:
Field Options:
New language for code-generation: Python
Model options:
Enum options:
Field types:
Field Options:
New language for code generation: GraphQL
Model Options:
Field types:
Field Options:
Added Model and Enum options per supported language.
TypeScript model options:
TypeScript Enum Options:
New language for code-generation: Protobuf
Field types:
Field options:
You can now share projects (and their models and enums) with other users. Each user can also have read/write or read permissions.
You can now graph the entire project and its models and enums. Right-click on the ERM (entity-relationship-model) image to save or print it.
Exactly what the title says... You can now sort Enum values.
New language for code-generation: Joi
Field types:
Field options:
You can now import your existing Typescript interfaces.
New language for code-generation: Mongoose
Field types:
Field options:
You can now view the code for a specific language instead of downloading it.
New language for code-generation: Go
Field Types:
Field Options:
New language for code-generation: Typescript
Field types:
Field options:
© 2026 Unified API Inc.