EVOID

Reference Runtime for Intent-Oriented Programming

Data declares what, runtime decides how.

What is IOP? • Quick Start • Adapters • Features • Docs • PyPI

What is IOP?

Intent-Oriented Programming — your data declares what it needs, the runtime handles how.

from evoid.native import create_service, on
from evoid import Intent, Level

app = create_service("my-api")

GET_USER = Intent(name="get_user", level=Level.STANDARD)

async def get_user(intent: Intent) -> dict:
    return {"id": 1, "name": "Alice"}

on(app, GET_USER, get_user)

Three intent levels control infrastructure automatically:

How It Works

EVOID is a runtime, not a framework. Adapters bridge the outside world:

External Event (HTTP, CLI, Telegram, WebSocket, ...)
        |
   Adapter converts event → Intent
        |
   Runtime executes Intent through Pipeline
        |
   Pipeline runs Processors (validate, authorize, audit, ...)
        |
   Result returned to Adapter → converted back to response

Adapters provide route decorators and event conversion. Services are Intent + handler registrations. Pipelines are processor chains chosen by intent level.

See How It Works for the full picture.

Quick Start

uv add evoid
evo init my-api && cd my-api
evo service run gateway    # already scaffolded, http://0.0.0.0:8000

evo service new api        # add another service (optional)
evo service run api        # http://0.0.0.0:8001

Or install optional engines:

evo install sqlite      # SQLite storage
evo install redis       # Redis cache
evo install pydantic    # Pydantic schema
evo install full        # Everything

Three Syntax Styles

All IOP underneath. Pick your style:

@route (function-based)

Route decorators come from the adapter. Switch adapters, same code:

from evoid.adapters.asgi import get, post
from evoid.web.route import Service

app = Service("my-api")

@get("/users/{user_id}")
async def get_user(user_id: int) -> dict:
    return {"id": user_id, "name": "Alice"}

@controller (class-based)

@GET, @POST mark routes. @Controller creates Intents from them:

from evoid.web.controller import Service, Controller, GET, POST

app = Service("my-api")

@Controller("/users")
class UserController:
    @GET("/{user_id}")
    async def get_user(self, user_id: int) -> dict:
        return {"id": user_id}

Native (full control)

Explicit Intent creation. Adapter-agnostic — works with any transport:

from evoid.native import create_service, on
from evoid import Intent, Level

app = create_service("my-api")

GET_USER = Intent(name="get_user", level=Level.STANDARD)

async def get_user(intent: Intent) -> dict:
    return {"id": 1, "name": "Alice"}

on(app, GET_USER, get_user)

Adapters

Each adapter converts its event type to Intents. Route decorators live in adapters because param extraction is adapter-specific:

See Adapters for details and examples.

Features

AI Agent Integration

from evoid import export_schemas
from evoid.adapters.mcp import create_mcp_server

schemas = export_schemas()
server = create_mcp_server("my-api")

Plugin System

evo plug install evoid-redis           # From PyPI
evo plug install git+https://...       # From git
evo plug search cache                  # Search PyPI
evo plug list                          # List installed

Configuration

Python (Recommended)

from evoid.config import config

app = config(
    service={"name": "my-api"},
    runtime={"adapter": "asgi", "port": 8000},
    engines={"storage": "redis"},
)

TOML

[service]
name = "my-api"

[engines]
storage = "redis"

Testing

from evoid.testing import tc
from myapp import GET_USER

def test_get_user():
    return tc(GET_USER, expect={"id": 1})

pytest tests/ -v                    # Run tests
pytest tests/ --evoid-webui         # With dashboard
pytest tests/ --evoid-inspect       # With pipeline traces

Project Structure

my-api/
  evoid.toml              # TOML config (or evoid_config.py)
  shared/                 # Shared code
  services/
    api/
      main.py             # Service code

Package Layout

evoid/
  core/          Intent, Pipeline, Context, Runtime, Events, MessageBus
  native/        IOP mother syntax (create_service, on)
  web/           @route and @controller syntax
  adapters/      asgi, cli, telegram, websocket, mcp, robyn
  engines/       storage, cache, auth, di, logger, metrics, schema, serializer
  processors/    Built-in processors (validate, authorize, etc.)
  config/        TOML + Python config loading
  testing/       pytest plugin + WebUI dashboard
  project/       Project scaffolding

Documentation

Full docs: https://evolvebeyond.github.io/EVOID/

Architecture: https://deepwiki.com/EvolveBeyond/EVOID (architecture documentation that updates faster, great for contributors)

Contributing

See CONTRIBUTING.md.

1. Fork → Branch → Commit → PR
2. Tests: pytest tests/ -v
3. Lint: ruff check evoid/

License

Apache 2.0

_{Built with IOP principles. Intent is the platform.}