--- title: "Deploy Agno" description: "A production-ready agent runtime with API, MCP, memory and observability." category: "AI/ML" url: https://railway.com/deploy/agno --- # Deploy Agno A production-ready agent runtime with API, MCP, memory and observability. **[Deploy Agno on Railway](https://railway.com/template/agno)** Machine-readable deploy manifest (JSON, validated by TemplateCI): https://railway.com/deploy/agno/manifest.json - **Creator:** INF Labs - **Category:** AI/ML ## Template content ### Postgres https://devicons.railway.app/i/postgresql.svg - **Image:** ghcr.io/railwayapp-templates/postgres-ssl:18 ### Agno https://app.agno.com/icons/agno_logo.svg - **Source:** agno-agi/agentos-railway - **Public domain:** Yes ## Documentation # Deploy and Host Agno AgentOS on Railway Agno AgentOS is a production-ready runtime for building, serving, and managing AI agent platforms. This Railway template deploys **AgentOS with PostgreSQL**, giving you persistent sessions, memory, knowledge, traces, REST APIs, MCP connectivity, scheduling, observability, and secure production deployment. ## About Hosting Agno AgentOS AgentOS is the runtime layer of the Agno ecosystem. It allows you to build AI agents using the Agno SDK, expose them through production APIs, connect them to external interfaces, and manage runtime state using your own infrastructure. This template uses PostgreSQL as the persistent storage backend for AgentOS data such as: * Sessions * Memory * Knowledge * Traces * Runtime state * Agent activity The AgentOS service is publicly accessible, while PostgreSQL communicates only through Railway private networking. ## Included Services * **Agno AgentOS** — Agent runtime, REST API, MCP server, scheduling, observability * **PostgreSQL** — Persistent storage for AgentOS data and traces * **Railway Private Networking** — Secure internal communication between AgentOS and PostgreSQL ## Common Use Cases * Build and host AI agents * Expose agents through REST APIs * Run multi-agent applications * Store persistent agent memory * Track sessions and execution traces * Build agent-powered backend services * Connect agents to MCP-compatible clients * Add human approval to agent workflows * Run scheduled agent jobs * Build internal AI platforms * Integrate agents with external tools and services * Deploy agent applications without managing your own server infrastructure ## Dependencies for Agno AgentOS Hosting This template includes: * Agno AgentOS * PostgreSQL Redis is not required for the standard AgentOS deployment. PostgreSQL is used as the primary persistent storage layer for sessions, memory, knowledge, traces, and other runtime data. ## AgentOS API AgentOS runs on port: ```text 8000 ``` After deployment, generate a public domain for the AgentOS service. A typical deployment URL looks like: ```text https://your-agentos-domain.up.railway.app ``` Interactive API documentation is available at: ```text https://your-agentos-domain.up.railway.app/docs ``` ## MCP Endpoint AgentOS can expose agents and tools through Model Context Protocol. The MCP endpoint is available at: ```text https://your-agentos-domain.up.railway.app/mcp ``` This allows compatible clients to connect to your AgentOS deployment. Potential integrations include: * ChatGPT * Claude * Claude Code * Codex * Cursor * Other MCP-compatible applications ## PostgreSQL Storage PostgreSQL runs as a private internal service. AgentOS connects to PostgreSQL using Railway service references. The database stores persistent AgentOS data including: * Agent sessions * Memory * Knowledge * Traces * Runtime metadata The PostgreSQL service does not need a public domain. Persistent storage is handled by the PostgreSQL Railway Volume. ## Important Environment Variables ### AgentOS `OPENAI_API_KEY` API key used by the default OpenAI-based agents and embeddings. Users can replace the default agent configuration with other supported model providers if required. `RUNTIME_ENV` Controls the AgentOS runtime environment. This template is intended to run with: ```text prd ``` for production deployments. `AGENTOS_URL` Public URL of the deployed AgentOS service. The template references the generated Railway public domain automatically. `JWT_VERIFICATION_KEY` Public verification key used by AgentOS production authentication. This value must be configured after connecting the deployment to the AgentOS Control Plane. `MCP_CONNECT_SECRET` Optional secret used for external MCP client connectivity. `PARALLEL_API_KEY` Optional API key used for web search and faster knowledge ingestion. `AGNO_TELEMETRY` Optional setting that controls anonymous Agno telemetry. ## PostgreSQL Connection AgentOS connects to PostgreSQL using: * `DB_HOST` * `DB_PORT` * `DB_USER` * `DB_PASS` * `DB_DATABASE` * `DB_DRIVER` These values reference the PostgreSQL Railway service directly. The expected database driver is: ```text postgresql+psycopg ``` ## Production Authentication Setup AgentOS production mode requires authentication configuration. After the first deployment: 1. Generate a public domain for the AgentOS service. 2. Open the AgentOS Control Plane. 3. Connect your deployed AgentOS instance. 4. Enable token-based authorization. 5. Copy the generated public JWT verification key. 6. Add the value to: ```text JWT_VERIFICATION_KEY ``` 7. Redeploy the AgentOS service. This step is important because production mode is designed to reject unsecured traffic when JWT verification is not configured. ## Networking The intended Railway architecture is: ```text Internet │ ▼ Railway HTTPS │ ▼ Agno AgentOS :8000 │ │ Railway Private Network ▼ PostgreSQL :5432 ``` Only AgentOS should be exposed publicly. PostgreSQL should remain private. ## Persistent Storage The AgentOS application service does not require its own persistent volume. Persistent application data is stored in PostgreSQL. The recommended persistence model is: ```text AgentOS │ ▼ PostgreSQL │ ▼ Railway Volume ``` This keeps the AgentOS runtime stateless and makes redeployments easier. ## After Deployment 1. Wait for AgentOS and PostgreSQL to become healthy. 2. Generate a public domain for AgentOS on port `8000`. 3. Keep PostgreSQL private. 4. Verify the AgentOS `/docs` endpoint. 5. Connect the deployment to the AgentOS Control Plane. 6. Configure `JWT_VERIFICATION_KEY`. 7. Redeploy AgentOS. 8. Configure the required model provider API key. 9. Test your first agent request. 10. Configure MCP connectivity if required. 11. Review sessions, traces, memory, and agent activity from the AgentOS platform. ## Security Notice For production deployments: * Keep PostgreSQL private * Protect all model provider API keys * Use production authentication * Configure `JWT_VERIFICATION_KEY` * Do not run public production deployments with unsecured development settings * Rotate exposed credentials immediately * Use scoped access tokens where possible * Review MCP connectivity before exposing external tools ## Agno AgentOS vs Alternatives | Feature | Agno AgentOS | Pydantic AI | Mastra | CrewAI | | ----------------------------- | -----------: | -------------: | -------: | -----------------------: | | Production agent runtime | ✅ | ✅ | ✅ | ✅ | | Built-in REST API | ✅ | Depends on app | ✅ | Depends on app | | Persistent PostgreSQL storage | ✅ | App-dependent | ✅ | App-dependent | | Agent memory | ✅ | ✅ | ✅ | ✅ | | MCP support | ✅ | ✅ | ✅ | ⚠️ | | Observability | ✅ | ✅ | ✅ | ⚠️ | | Scheduling | ✅ | Depends | Depends | Depends | | Multi-user / RBAC | ✅ | App-dependent | ⚠️ | Commercial control plane | | Web management platform | ✅ | ⚠️ | ✅ Studio | ⚠️ | | Railway-oriented deployment | ✅ | ⚠️ | ⚠️ | ⚠️ | | Self-hosted runtime | ✅ | ✅ | ✅ | ✅ | Agno AgentOS is particularly useful when you want a complete agent runtime with persistent storage, production APIs, MCP support, observability, and infrastructure ownership. ## Why Deploy Agno AgentOS on Railway? Railway is a singular platform to deploy your infrastructure stack. Railway will host your infrastructure so you don't have to deal with configuration, while allowing you to vertically and horizontally scale it. Deploying Agno AgentOS on Railway provides: * Automatic HTTPS domains * Private PostgreSQL networking * Persistent agent data * Simple environment variable management * Easy redeployments and upgrades * Integrated logs and monitoring * Production-ready REST APIs * MCP connectivity * Agent memory and knowledge storage * A clean two-service architecture This template provides a practical way to run your own AgentOS deployment with persistent PostgreSQL storage and production-focused infrastructure. ## Similar templates - [Chat Chat](https://railway.com/deploy/-WWW5r) — Chat Chat, your own unified chat and search to AI platform. - [stella](https://railway.com/deploy/stella) — Self-host stella with web, API, Postgres, Redis, and object storage. - [Hermes Agent | OpenClaw Alternative with Dashboard](https://railway.com/deploy/hermes-agent-or-openclaw-alternative-wit) — Self-Hosted Hermes AI Agent for Telegram, Discord & Slack Open this page in a browser: https://railway.com/deploy/agno