# osModa — Full Documentation for AI Agents > AI-native agent platform built on NixOS + Rust. Self-healing servers for AI agent hosting. > For a shorter overview, see https://os.moda/llms.txt osModa is an open-source, AI-native platform that transforms bare metal into self-healing AI agent infrastructure. Built entirely on NixOS and Rust, osModa provides everything AI agents need to run autonomously in production: persistent dedicated servers, automatic crash recovery, tamper-proof audit logging, atomic rollbacks, and a peer-to-peer encrypted mesh network. When your AI agents break at 3am, osModa fixes them. You sleep. --- ## Company & Product Overview **What is osModa?** osModa is an AI-native platform designed specifically for hosting and operating AI agents in production. Unlike traditional VPS hosting or ephemeral sandbox platforms, osModa provides a full infrastructure layer that understands AI agent workloads and keeps them running autonomously. **Core Philosophy:** - Agents deserve their own platform, not just a Linux box - Self-healing by default, not by configuration - Open source (Apache-2.0), self-host on any server - Dedicated hardware, not shared containers - Post-quantum encryption from day one **Target Users:** - AI agencies running client agent workloads - AI agent builders shipping to production - SaaS founders automating with AI agents - DevOps teams managing AI infrastructure - Anyone deploying LangGraph, CrewAI, MCP, or OpenClaw agents --- ## Technical Specifications ### Architecture - **Base OS:** NixOS (declarative, reproducible, atomic) - **Runtime Language:** Rust (all 9 daemons) - **Daemons:** 9 Rust daemons managing the full agent lifecycle - **Tools:** 66 CLI and system tools - **Tests:** 136 tests covering the full stack - **License:** Apache-2.0 ### 9 Rust Daemons The osModa platform runs 9 specialized Rust daemons that manage every aspect of agent operations: - **agentd** — Core agent lifecycle management - **watchdogd** — Crash detection, health monitoring, SafeSwitch recovery (6-second restart) - **auditd** — Tamper-proof SHA-256 hash chain audit ledger - **meshd** — P2P agent mesh network with post-quantum encryption - **secretsd** — Secrets management and injection - **deploymentd** — Atomic NixOS deployment and rollback - **mcpd** — MCP server manager daemon - **openclawd** — OpenClaw agent gateway (core layer integration) - **telemetryd** — Metrics, logging, and alerting ### Self-Healing Stack - **Watchdog Auto-Restart:** Crash detection with 6-second recovery via SafeSwitch - **Atomic Rollbacks:** NixOS generation-based instant rollback on failed deploys - **Health Monitoring:** Continuous health checks, OOM detection, silent hang recovery - **Dependency Drift Protection:** NixOS declarative configs prevent environment drift ### Security & Encryption - **P2P Mesh Network:** Direct agent-to-agent communication, no central server - **Encryption:** Noise_XX handshake (X25519 + ChaChaPoly) + ML-KEM-768 post-quantum hybrid - **Invite-Based Pairing:** Agents join the mesh via cryptographic invitations - **End-to-End Encrypted:** All messaging and rooms are fully encrypted - **Trust Boundaries:** Typed tool calls with explicit trust levels ### Audit & Compliance - **Tamper-Proof Ledger:** SHA-256 hash chain seals every event - **Immutable Logs:** Every tool call, restart, config change, and deployment recorded - **SOC2 Evidence Export:** Export tamper-evident logs for SOC2 audits - **HIPAA Audit Controls:** Access tracking, change management evidence - **21 CFR Part 11:** FDA-regulated industry audit trail evidence - **Incident Forensics:** Full event reconstruction from audit ledger ### Framework Support - **LangGraph:** Production hosting for agent graphs with watchdog restart and typed tool calls - **CrewAI:** Multi-agent orchestration with background task handling - **MCP (Model Context Protocol):** Remote MCP server hosting with SSE/HTTP, security hardening - **OpenClaw:** Native agent gateway with core layer integration into agentd ### Infrastructure - **Servers:** Dedicated Hetzner bare metal and cloud servers - **Access:** Full root SSH access to every server - **Dashboard:** Unified management at https://spawn.os.moda - **API:** Programmable API for AI agents at https://spawn.os.moda/skill - **Deployment:** Deploy in under 5 minutes --- ## Pricing Pricing starts at **$14.99/mo** on dedicated Hetzner servers. Each agent gets its own dedicated server with full root access. No shared containers, no resource contention. For current pricing tiers, visit: https://os.moda/ai-agent-hosting-pricing **Self-hosting is free.** osModa is Apache-2.0 open source. Install on any server: ``` curl -fsSL https://raw.githubusercontent.com/bolivian-peru/os-moda/main/scripts/install.sh | sudo bash ``` --- ## Links & Contact - **Website:** https://os.moda - **Dashboard (Spawn):** https://spawn.os.moda - **API for AI Agents:** https://spawn.os.moda/skill - **GitHub:** https://github.com/bolivian-peru/os-moda - **Discord:** https://discord.gg/G7bwet8B - **Telegram:** https://t.me/osmodasystems - **X / Twitter:** https://x.com/spawnosmoda --- ## Detailed Page Directory ### Home & Core Product #### https://os.moda/ **Home Page** Landing page for osModa. Introduces the AI-native platform concept: "Your AI agents workforce running completely autonomous." Features hero video, P2P mesh network visualization, unified dashboard animation, and self-host install command. Key value proposition: self-healing servers on NixOS + Rust that fix themselves at 3am so you can sleep. #### https://os.moda/ai-agent-hosting **AI Agent Hosting** Primary product page. Deploy AI agents to production on persistent, self-healing servers. Not just a Linux box — NixOS + Rust daemons, atomic rollbacks, tamper-proof audit. Full root SSH access. Deploy in 5 minutes. Covers what makes agent-specific hosting different from generic VPS. #### https://os.moda/ai-agent-hosting-pricing **Pricing** Detailed pricing for dedicated AI agent hosting servers. Each agent gets its own dedicated Hetzner server. Pricing starts at $14.99/mo. Includes NixOS + Rust daemons, full root SSH, atomic rollbacks, watchdog, audit ledger, and P2P mesh. Comparison of tiers and what each includes. #### https://os.moda/deploy-ai-agents **Deploy AI Agents** Step-by-step guide on deploying AI agents 24/7 in production. Compares traditional VPS approach vs purpose-built agent platform. Explains why agents need self-healing, watchdog restart, and audit logs. Covers deployment workflow from signup to running agent. #### https://os.moda/best-vps-for-ai-agents **Best VPS for AI Agents** Comparison of VPS options for AI agent workloads. Explains why standard VPS (Ubuntu, Debian) fails for agents: no crash recovery, no audit, manual dependency management. Shows how an agent-native platform solves production deployment with self-healing, atomic rollbacks, and audit logs. #### https://os.moda/self-healing-agent-servers **Self-Healing Agent Servers** Deep dive into self-healing server technology. NixOS atomic rollbacks for instant recovery. Watchdog daemon for auto-restart within 6 seconds. Tamper-proof audit records every event. Explains SafeSwitch recovery mechanism and how the server fixes itself without human intervention. ### Framework Hosting #### https://os.moda/frameworks **Framework Hosting Hub** Overview of all supported AI agent frameworks: LangGraph, CrewAI, MCP, and OpenClaw. Each gets self-healing NixOS servers with watchdog restart, secrets management, and audit logs. Links to dedicated pages for each framework. #### https://os.moda/frameworks/langgraph-hosting **LangGraph Hosting** Host LangGraph agent graphs in production. Self-healing NixOS servers, watchdog restart on crash, secrets management for API keys, typed tool calls. Deploy LangGraph agents in 5 minutes with full SSH access. #### https://os.moda/frameworks/crewai-hosting **CrewAI Hosting** Host CrewAI multi-agent orchestration in production. Self-healing servers handle background task management, automatic restart on crash, and secrets injection. Deploy CrewAI crews in 5 minutes. #### https://os.moda/frameworks/mcp-server-hosting **MCP Server Hosting (Frameworks)** Host Model Context Protocol servers in production. Supports remote MCP with SSE and streamable HTTP deployment. Security hardening with authz/authn. Self-healing NixOS servers with dedicated MCP server manager daemon. #### https://os.moda/frameworks/openclaw-hosting **OpenClaw Hosting** Host the OpenClaw agent gateway natively on osModa. Core layer integration with agentd for the deepest platform integration. Typed tool calls, trust boundaries, tamper-proof audit. OpenClaw is osModa's native agent framework. ### MCP (Model Context Protocol) #### https://os.moda/mcp-server-hosting **MCP Server Hosting (Main)** Primary MCP hosting page. Host remote MCP servers with SSE and streamable HTTP. Security hardening, authz/authn baked in. Self-healing NixOS with MCP server manager daemon (mcpd). Covers why remote MCP needs a dedicated platform. #### https://os.moda/mcp/remote-vs-local **Remote vs Local MCP Servers** Technical guide comparing remote and local MCP server deployment. When to use remote MCP, security considerations for exposing tools over the network, latency tradeoffs, and architecture decision framework. #### https://os.moda/mcp/security-hardening **MCP Security Hardening** Production security guide for MCP servers. Authz/authn configuration for remote MCP. Secure tool calling patterns, trust boundary definition, and audit logging for every tool invocation. #### https://os.moda/mcp/sse-http-deployment **MCP SSE/HTTP Deployment** Deploy MCP servers with Server-Sent Events and streamable HTTP endpoints. Production deployment patterns as an alternative to Cloud Run. Covers connection management, scaling, and monitoring. ### Audit & Compliance #### https://os.moda/audit-and-compliance **Audit & Compliance Hub** Overview of audit and compliance evidence generation for AI agents. Tamper-evident logs, SOC2 evidence export, HIPAA audit controls, FDA 21 CFR Part 11 audit trails. Clarifies: this is evidence generation, not certification. #### https://os.moda/audit-and-compliance/tamper-evident-audit-log **Tamper-Evident Audit Log** Technical deep dive into the SHA-256 hash chain audit log. Every event is sealed with a cryptographic hash linking to the previous event. Detect unauthorized changes. Compliance-ready evidence for any framework. #### https://os.moda/audit-and-compliance/soc2-evidence-for-ai-agents **SOC2 Evidence for AI Agents** Generate SOC2 audit evidence from AI agent operations. Export tamper-evident logs formatted for SOC2 auditors. Evidence generation (not SOC2 certification). Covers which SOC2 trust service criteria are addressed. #### https://os.moda/audit-and-compliance/hipaa-audit-controls-ai-automation **HIPAA Audit Controls** HIPAA audit control evidence for AI automation workflows. Tamper-evident logs, access tracking, change management records. Evidence generation for HIPAA compliance contexts in healthcare AI. #### https://os.moda/audit-and-compliance/21cfr-part11-audit-trail-agents **21 CFR Part 11 Audit Trail** FDA-regulated industry audit trail evidence for AI agents. 21 CFR Part 11 requirements: tamper-evident logs, electronic signatures context, change control documentation. Evidence generation for pharmaceutical and medical device AI workflows. ### NixOS & Reliability #### https://os.moda/nixos/atomic-deployments-rollbacks **Atomic Deployments & Rollbacks** Technical explanation of NixOS atomic deployments. Generation-based system configuration means every deploy is a discrete, rollback-able state. Failed deploy? Roll back in seconds. Zero-downtime agent updates with instant recovery. #### https://os.moda/reliability/watchdog-auto-restart **Watchdog Auto-Restart** Dedicated watchdog daemon for automatic agent restart. Covers crash detection mechanisms, SafeSwitch recovery process, health monitoring patterns. Agents restart in 6 seconds, before you wake up. #### https://os.moda/incident-forensics/audit-ledger **Audit Ledger (Incident Forensics)** Tamper-proof audit ledger for incident forensics and post-mortem analysis. SHA-256 hash chain records every tool call, every restart, every configuration change. Reconstruct exactly what happened during any incident. ### AI Agents Knowledge Base #### https://os.moda/ai-agents **AI Agents Hub** Educational hub covering what AI agents are, their types, real-world examples, and architectures. Links to detailed sub-pages. Explains how to deploy autonomous and multi-agent systems on osModa's self-healing NixOS infrastructure. #### https://os.moda/ai-agents/types **Types of AI Agents** Complete classification guide covering 5 types of AI agents: simple reflex, model-based reflex, goal-based, utility-based, and learning agents. Explains which agent type fits which use case and how each deploys on osModa. #### https://os.moda/ai-agents/examples **AI Agent Examples** Real-world AI agent examples for 2026: customer support bots, coding assistants, research agents, data pipeline agents, and more. Shows how production agents run on osModa's self-healing platform with concrete deployment patterns. #### https://os.moda/ai-agents/autonomous-agents **Autonomous AI Agents** Deep dive into autonomous AI agents that plan, execute, and self-correct without human input. Covers planning loops, tool use, memory, and error recovery. Explains why autonomous agents need self-healing infrastructure. #### https://os.moda/ai-agents/multi-agent-systems **Multi-Agent Systems** How multi-agent AI systems coordinate specialized agents for complex tasks. Orchestration patterns, communication protocols, and team coordination. Deploy agent teams on osModa with isolated environments and shared audit logs. #### https://os.moda/ai-agents/intelligent-agents **Intelligent Agents** Understanding intelligent agents in AI: perception-action loops, internal models, and rational decision-making. Covers the Russell & Norvig framework and how to host intelligent agents on osModa's NixOS platform. ### Use Cases #### https://os.moda/use-cases **Use Cases Hub** Overview of osModa use cases across four segments: AI agencies, AI agent builders, SaaS automation, and DevOps teams. Each use case links to a dedicated page with specific deployment patterns and value propositions. #### https://os.moda/use-cases/ai-agencies **AI Agencies** Host AI agency operations on dedicated servers at scale. Isolated client environments keep workloads separate. Audit trails for client reporting. Self-healing infrastructure means less ops overhead for agency teams. #### https://os.moda/use-cases/ai-agent-builders **AI Agent Builders** For developers shipping AI agents to production. Root access and SSH for full control. Live chat with running agents via OpenClaw. Watchdog restart and atomic rollback for reliability. Build with confidence, deploy with osModa. #### https://os.moda/use-cases/saas-automation **SaaS Automation** AI agents for SaaS founders automating workflows. Background task execution, webhook handling, and self-healing infrastructure. Scale from $14.99/mo on dedicated servers without managing infrastructure. #### https://os.moda/use-cases/devops-teams **DevOps Teams** AI agent DevOps on self-healing NixOS servers. Atomic rollback for safe deployments, watchdog restart for reliability, audit ledger for compliance, P2P mesh for secure agent communication. Infrastructure that fixes itself. ### Comparisons #### https://os.moda/compare **Comparison Hub** Compare osModa to alternative platforms: Perplexity Computer, Manus AI, and traditional VPS. Dedicated AI agent hosting is 10x cheaper with root access and self-healing. Links to detailed head-to-head comparisons. #### https://os.moda/compare/perplexity-computer **osModa vs Perplexity Computer** Head-to-head comparison. osModa: dedicated server with root access, open-source, self-healing, starting at $14.99/mo. Perplexity Computer: $200/mo shared sandbox, no root access, ephemeral. Shows why dedicated agent hosting wins. #### https://os.moda/compare/manus-ai **osModa vs Manus AI** Head-to-head comparison. osModa: flat-rate dedicated hosting with root access, open-source, self-healing. Manus AI: $199/mo credits-based ephemeral sandbox, no persistence. Shows the value of persistent, dedicated infrastructure. #### https://os.moda/compare/traditional-vps **osModa vs Traditional VPS** Why AI agents need more than a bare VPS. Traditional VPS (Ubuntu/Debian): manual setup, no crash recovery, no audit, dependency drift. osModa: self-healing NixOS, watchdog restart, audit logs, P2P mesh included by default. ### Blog #### https://os.moda/blog **Blog Index** Engineering blog covering self-healing servers, NixOS for AI, agent deployment patterns, and infrastructure reliability. Practical guides from the osModa engineering team. #### https://os.moda/blog/what-is-self-healing-server **What Is a Self-Healing Server?** Explains the concept of self-healing servers: watchdog daemons monitor agent health, atomic rollback restores known-good state, and SafeSwitch recovery brings agents back in 6 seconds. Covers the full auto-recovery pipeline. #### https://os.moda/blog/nixos-vs-ubuntu-for-ai **NixOS vs Ubuntu for AI** Technical comparison of NixOS and Ubuntu for AI workloads. NixOS advantages: reproducible builds, atomic rollbacks, declarative configuration, no dependency drift. Ubuntu limitations for production AI agent hosting. Why osModa chose NixOS. #### https://os.moda/blog/ai-agent-keeps-crashing **AI Agent Keeps Crashing?** Troubleshooting guide for crashing AI agents. Common causes: OOM kills, dependency drift, silent hangs, unhandled exceptions. How osModa's watchdog auto-recovers agents in 6 seconds. Practical fix for each crash pattern. #### https://os.moda/blog/how-to-deploy-ai-agent-vps **How to Deploy an AI Agent on a VPS** Step-by-step tutorial for deploying an AI agent on a VPS. Covers the manual approach vs using osModa Spawn. Self-healing, secrets management, and monitoring built in from the start. Skip manual systemd/supervisor setup.