Open IntentModeling Language

An open standard for AI-driven development. Declare what you want to build, and let AI agents generate the code — deterministically, safely, and repeatedly.

View On GitHub How It Works

What is OIML?

OIML (Open Intent Modeling Language) is an open standard that enables declarative, AI-driven development.

Instead of writing code directly, you declare spec-driven intents. AI agents (Cursor, Claude, GitHub Copilot) read these intents and generate production-ready code — deterministically and safely.

Traditional: You write code.

Vibe Coding: You vibe. AI writes iffy code.

OIML: You declare intent. AI writes good code.

Get Started

OIML is open source and ready to use.

View On GitHub View Docs

Quick Start

# Install the CLI

npm install -g @oiml/cli

# Start local MCP Server

oiml mcp

# Initialize OIML in a project

oiml init

# Create your first intent

oiml create FEAT-1

Key Features

Built for AI agents from the ground up

Declarative Development

Describe what you want, not how to build it

Standardized Schema

All intents follow a consistent, versioned structure

Deterministic Generation

Same intent always produces the same code

Framework Agnostic

Use your preferred stack

LLM Agnostic

Works with Cursor, Claude, Codex, and more

Intermediate Representation

Intents transform to IR for framework-specific generation

Version Control Friendly

Intent files live alongside your code in Git, leaving a clean audit trail

Migration Safety

Automatic database migrations with rollback support

Integrations

Drop in full capabilities with a single intent

OIML vs Traditional Development

Traditional Development	OIML
Write code manually	Declare intent, AI generates code
AI guesses from context	AI executes from standardized schema
Inconsistent AI output	Deterministic, repeatable generation
Code diffs only	Intent files + versioned schemas
Manual migration scripts	Automatic migration generation
No standard for AI collaboration	Open standard designed for AI agents

How It Works

OIML transforms declarative intents into production-ready code through a simple workflow.

# intent.yaml

version: 0.1.0
intents:
  - type: add_entity
    entity: User
    fields:
      - name: id
        type: uuid
        required: true
      - name: email
        type: string
        unique: true
      - name: name
        type: string

Supported Intent Types

Declare what you want to build

Data

•add_entity— Create database models
•add_field— Add fields to entities
•add_relation— Define relationships
•remove_entity/remove_field
•rename_entity/rename_field

API

•add_endpoint— Create REST endpoints
•update_endpoint— Modify responses

Capability

•add_capability— Add auth, file upload, SSE, websockets

UI

•add_component— Create UI components (coming soon)

Use Cases

Rapid Prototyping

Declare your data models and API endpoints. OIML generates the entire backend in minutes — with migrations, types, and CRUD operations.

AI-First Development

Build apps where AI agents handle feature development. Your team reviews intents instead of code diffs.

Maintainable Systems

Intent files serve as living documentation. AI agents can understand your architecture and safely extend it over time.

Why OIML?

AI agents are powerful, but they need structure.
OIML provides the standard that makes AI-driven development reliable.

No guessing — AI executes from explicit schemas, not context
Deterministic output — same intent always produces same code
Framework agnostic — IR transforms to any framework
Open standard — not locked to any vendor or tool