Skip to content

ALMAAutonomous Language Model Architecture

Infrastructure as Conversation - Manage your infrastructure using natural language

ALMA Terminal Demo

Quick Example ​

Conversational Infrastructure Management:

bash
# 1. Clone & Configure
git clone https://github.com/fabriziosalmi/alma.git
cd alma
cp .env.example .env

# 2. Start ALMA Stack (API + Web + Redis)
docker compose up -d

# 3. Chat with your Infrastructure
# Open http://localhost:3000

Core Features ​

Quick Wins (v0.8.5 - v0.8.8) ​

  • Docker Engine: Native support for managing containers with robust logging.
  • Live Dashboard: Visualize system resources (VMs, Containers) in real-time.
  • Resilience: Implemented Circuit Breaker and Retry patterns for core system stability.
  • Test Coverage: Significantly improved unit test coverage (~30%) for Core, Engines, and API.
  • The Council: Multi-Agent system where specialized AI agents (Architect, SecOps, FinOps) debate infrastructure decisions.
  • GraphQL API: Flexible data querying (/graphql) powered by Strawberry.

LangGraph Orchestration ​

ALMA uses a state machine to handle complex deployment workflows, error recovery, and verification loops.

MCP Native ​

Built on the Model Context Protocol, allowing integration with Claude Desktop, Cursor, and other AI agents.

Proxmox Engine ​

Engine with task-aware waiting (no more race conditions), dual SSH/API modes, and automatic template management.

Security ​

Built-in security auditing, compliance checking, and best practices enforcement for production deployments.

Scalable Architecture ​

Horizontally scalable design with support for distributed deployments and high-throughput workloads.

Roadmap ​

Intelligent Orchestration (In Progress) ​

  • Multi-Agent Architecture: Specialized agents for Architecture, Security, and Cost.
  • Proactive Security: Pre-deployment vulnerability scanning in IPRs.

Predictive Operations (Planned) ​

  • Anomaly Detection: ML-based resource usage analysis.
  • Capacity Forecasting: Historical data-driven scaling predictions.
  • Auto-Remediation: Automated recovery runbooks.

Universal Abstraction (Planned) ​

  • Cloud Portability: Abstracted blueprints for AWS/Azure/GCP migration.
  • Hybrid Deployment: Unified management for On-Prem and Cloud.

Getting Started ​

  1. Quick Start Guide - Get up and running quickly
  2. Architecture Overview - Understand ALMA's design
  3. API Reference - Complete API documentation
  4. Production Deployment - Deploy to production

Community & Support ​

Released under the Apache 2.0 License.