fastloom-sdk

Fastloom – The Open Foundation for Building Event-Driven Services

Fastloom is a lightweight, batteries-included foundation for building modern backends. Define your settings, schemas, and endpoints; Fastloom wires up the rest: FastAPI, Mongo (Beanie), Rabbit (FastStream), metrics/traces/logs/errors, and more.

Think of it as the glue for your stack: web, messaging, caching, DB, observability, and integrations with best-in-class tools.

---

Why Fastloom

• No boilerplate: minimal scaffolding/templating; most wiring is handled inside the library.
• Composable: opt into only what you need via extras (fastapi, rabbit, kafka, mongo, redis, mcp, celery, openai).
• Pydantic-first: type-safe models, validators, and clear input/output contracts.
• Multi-tenant by design: tenant context flows through DI and storage.
• AuthN/Z via DI: OIDC token introspection and pluggable PDP (ABAC/RBAC/ReBAC) hooks.
• Event-driven ready: publish/subscribe with routing keys and health.
• Observability-native: metrics, traces, logs from day one.
• Self-hostable: production parity with a cloud/aaS setup.

---

Integrated Services (the platform)

Fastloom plugs into a family of self-hostable services:

• IAM → OIDC/SSO, authN/Z, RBAC/ABAC/ReBAC.
• Notify → realtime notifications, Pusher-compatible API.
• Pulse → user activity + event tracking with OpenTelemetry hooks.
• File → object storage on MinIO (S3-compatible).
• Finance, Subscription, SMS/Email, Meet, Persona → optional services you can wire in.

Each service is:

• self-hostable (Docker Compose or Helm),
• BaaS-available.

---

Quick start

# Install fastloom with the extras you need
poetry add fastloom -E fastapi -E mongo -E rabbit

A minimal service is two files at the project root — settings.py and app.py — plus a tenants.yaml for defaults:

# settings.py
from fastloom.db.settings import MongoSettings
from fastloom.launcher.settings import LauncherSettings
from fastloom.observability.settings import ObservabilitySettings
from fastloom.settings.general import BaseGeneralSettings
from fastloom.signals.settings import RabbitmqSettings


class Settings(
    BaseGeneralSettings,
    LauncherSettings,
    MongoSettings,
    RabbitmqSettings,
    ObservabilitySettings,
): ...

# app.py
from fastapi import APIRouter

from fastloom.launcher.schemas import App

from my_service import models, signals

router = APIRouter()


@router.get("/ping")
async def ping() -> dict[str, str]:
    return {"pong": "ok"}


app = App(
    routes=[(router, "", "Health")],
    models_module=models,
    signals_module=signals,
)

# tenants.yaml
default:
    ENVIRONMENT: development
    PROJECT_NAME: my_service
    MONGO_URI: mongodb://localhost:27017
    MONGO_DATABASE: my_service
    RABBIT_URI: amqp://guest:guest@localhost:5672/

Run with the bundled CLI (registered as launch via [project.scripts]):

launch

See docs/quickstart.md for a fuller walkthrough.

---

What you get out of the box

• App orchestrator (fastloom.launcher)
  • Discovers your routes, models, signals, and healthchecks from app.py / settings.py
  • Exposes settings and health endpoints (public toggle via LauncherSettings.SETTINGS_PUBLIC)
• FastAPI-native
  • Dependency-injected request/tenant context and guards
  • Clear routing, OpenAPI, and dependency injection patterns
• Auth & Access
  • DI-based guards with OIDC / OAuth2 token introspection
  • Optional IAM sidecar for introspection + ACL
• Multi-tenancy
  • Tenant-aware DI context across web, DB, and messaging
  • Per-tenant settings endpoint backed by DB + cache, with tenants.yaml defaults
• Database layer (MongoDB via Beanie)
  • Created/updated mixins, pagination utilities, typed helpers
  • Auto model discovery for DB init via App.models_module
• Signals / Messaging (Rabbit via FastStream)
  • Event-driven publish/subscribe with retries and DLX-based backoff
  • Subscriber wiring and healthchecks
  • Auto-streamed BaseDocumentSignal Beanie models
• Observability
  • OpenTelemetry distro + OTLP exporter, Logfire, Sentry (errors + profiling)
• I18N
  • Exception handler and template utils with Babel/Jinja2
• Healthchecks
  • Automatic app/DB/messaging/cache checks + system routes
• MCP
  • Optional FastMCP mount with bearer auth forwarding
• Pydantic-native schemas and validators
  • Schema In/Out validation for request/response contracts
  • Common types and validators (fastloom.types)

Dive deeper in the docs below.

---

Documentation