Deployment Overview

LeafMesh deployment requires Python, Redis, and LLM API access. This section covers requirements, deployment options, and operational considerations.

System Requirements

Component	Minimum	Recommended
Python	3.11	3.11 – 3.14
Redis	6.0+	7.0+
Memory	4GB	8GB+
LLM API Access	1 provider	2+ providers

Dependencies

Required:

aiohttp — Async HTTP client
redis — Redis client
pydantic — Configuration validation
apscheduler — Agent scheduling

Optional:

fastapi, uvicorn — API services
pyyaml — YAML configuration loading

Observability dependencies are bundled by LeafMesh — no extra packages required.

Deployment Options

Option	Best For
Single process	Development, small deployments (one instance, no workers mode)
Managed containers	Production via [`leafmesh-deploy`](cloud-deploy) — App Runner / Container Apps from your project's Dockerfile
Docker / docker-compose	Reproducible deployments, CI/CD (every scaffolded project ships a Dockerfile + docker-compose.yml)
Kubernetes	Existing clusters; auto-scaling, high availability
Multi-instance	Horizontal scaling — requires `LEAFMESH_WORKERS_MODE=1` and a shared Redis ([Scaling](scaling))

Running more than one instance? Set LEAFMESH_WORKERS_MODE=1 on every instance and point them at the same Redis. See Scaling for the guarantees this enables.

Minimal Deployment

# app.py
import asyncio
from leafmesh import LeafMesh

leafmesh = LeafMesh.from_store()

async def main():
    await leafmesh.start()
    # Application logic...
    await leafmesh.stop()

asyncio.run(main())

With Redis running locally:

# Start Redis
redis-server

# Run the application
python app.py

With API Server

import asyncio
from leafmesh import LeafMesh

leafmesh = LeafMesh.from_store()

async def main():
    await leafmesh.start()
    # API server runs on port 18820 by default
    # Includes all service endpoints

asyncio.run(main())

Network Requirements

Redis: localhost:6379 (default) or cluster addresses
LLM APIs: HTTPS to provider endpoints (api.openai.com, api.anthropic.com, etc.)
API server: Port 18820 (configurable) for external access

Next Steps

Production Setup — Production configuration
Docker Deployment — Container deployment
Kubernetes Deployment — Kubernetes orchestration
Scaling — Scaling strategies

LeafMesh — Deployment overview