Skip to content

About

Experimental offline no-AI compiler for deterministic applications written in controlled natural English.

Topics

Resources

Contributing

Security policy

Stars

1 star

Watchers

0 watching

Forks

Repository files navigation

IntentLang

status node version

What if software requirements were executable?

IntentLang turns application intent into the program itself. Data models, relationships, business rules, workflows, permissions, and interface behavior are written in controlled English and compiled into a working full-stack system.

Software intent normally passes through requirements, tickets, architecture, code, database schemas, APIs, security rules, and tests. Every translation can lose meaning. IntentLang collapses that chain: the source stays readable enough to review as intent and precise enough to execute as software.

This is not prompt-to-code and not arbitrary natural language. There is no hidden model guessing what the author meant. Accepted sentences have defined semantics; unsupported or ambiguous instructions fail explicitly.

The English is the source code. The compiler is the authority.

IntentLang controlled English flowing through a deterministic compiler into a complete full-stack system

Warning

IntentLang v0.8.0-alpha.0 is experimental. It is not production-ready or intended for sensitive data, and it intentionally rejects unsupported or ambiguous instructions.

A language for intent, not implementation noise

application Todo
authentication uses User identified by email

role Member

a User has a required name as text
a User has a required unique email as text length between 1 and 320
a Task has a required title as text length between 1 and 200
a Task has a done as boolean default false
each Task belongs to a User as owner on delete restrict

action complete a Task
  require done is false otherwise "Task is already complete"
  set done to true

allow Member to read User where self
allow Member to create Task with owner as self
allow Member to read Task where owner is self
allow Member to update Task where owner is self
allow Member to run complete on Task where owner is self

Those statements are not comments, documentation, or an AI prompt. They define the application's data, ownership, authorization, and state transition. The compiler turns them into a typed application model and generates:

  • Responsive browser UI with forms, tables, relationships, and workflow actions
  • Node.js backend and validated REST API
  • SQLite schema, constraints, foreign keys, and migration plans
  • Authentication, role-based authorization, owner scoping, CSRF protection, idempotency, optimistic concurrency, and audit records
  • Canonical source, semantic fingerprints, manifests, and source-to-artifact traceability

Proof: real systems built from IntentLang

Each case study links to the controlled-English source, complete generated frontend/backend/database snapshot, and real desktop and mobile screenshots. Explore the full example gallery.

Choose your authoring path

  • Code is the direct IntentLang IDE for writing the formal business language with autocomplete, diagnostics, formatting, and a live application model. Start here when editing an existing .intent file.
  • App Builder is a guided path for turning a finite set of supported business-app descriptions into reviewable IntentLang source. It reports unsupported requests instead of silently improvising. See Describe App mode.
  • Visual Language is a separate controlled-English grammar for pages, styling, layout, animation, forms, and safe browser interactions. See the visual language guide.

More than business applications

IntentLang also includes an experimental visual/page language:

Show Hello, world!
Make the text large, blue and bold
Make the background white
Move the text from left to right over 3 seconds

See the visual language guide and example gallery.

Try it

Requires Node.js 24+.

git clone https://github.com/sethiramicrosoft/intentlang.git
cd intentlang
npm install
npm run sample:studio

Open http://127.0.0.1:3211 if Studio does not open automatically.

To inspect the compiler pipeline from the command line:

npm run sample:check
npm run sample:format
npm run sample:compile
npm run sample:generate

See Getting started for installation, generated-app startup, and safe account bootstrap instructions.

What works today

  • Deterministic parsing, formatting, compilation, and generation
  • Business entities, relationships, validation, roles, scoped permissions, policies, workflows, and state machines
  • Exact decimals, money, dates, typed functions, records, collections, and authorization-aware queries
  • Transactions, rollback, idempotency, typed external contracts, bounded retries/timeouts, and capability-scoped escape hatches
  • Modules, imports, namespaces, dependency locks, and semantic compatibility
  • Studio, App Builder wizard, LSP, REPL, debugger, conformance runner, semantic manifest validator, and backend parity checker
  • Visual pages with validated HTML/CSS, forms, layout, and safe interactions

This is controlled English, not arbitrary natural-language programming. Unsupported meaning is rejected rather than guessed. See Project status and limitations.

Documentation

Goal Read
Install and run IntentLang Getting started
Understand every Studio feature Studio feature guide
Learn the business language Business language specification
Build visual pages Visual language guide
Use Studio and the App Builder Describe App mode
Review the flagship application LaunchOps case study
Review another generated full-stack app Issue Tracker case study
Look up commands CLI reference
Understand safety and limitations Project status and limitations
Report or evaluate a vulnerability Security policy
Inspect formal language assurance Normative specification index
Review language-change requirements Language governance
Contribute Contributing guide

The complete categorized list is in the documentation index, but every primary path is linked directly above.

Project status

IntentLang is currently v0.8.0-alpha.0. Generated applications are for experimentation and evaluation, not production deployment.

No software license has been selected yet. Reuse rights are not granted beyond GitHub's viewing and forking terms.

About

Experimental offline no-AI compiler for deterministic applications written in controlled natural English.

Topics

Resources

Contributing

Security policy

Stars

1 star

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages