# Monk > Monk is the operator for the software your agents write. Your coding agent writes it; Monk ships it and keeps it running — on your cloud, on your bill. Monk builds fresh infrastructure from your codebase. Your existing deployment stays untouched. Test the new environment before you switch. The orchestrator automatically recovers workloads within configured policies. Watcher adds AI diagnosis; its proposed fixes need your approval. Resources run in your cloud account. You pay your cloud provider directly, separately from Monk and your coding agent. ## Explore Monk - [Pricing](https://monk.io/pricing): Choose product controls and managed capacity; cloud and coding-agent costs remain separate. - [Capsules](https://monk.io/capsules): Give branches separate application environments with reviewed resource and data boundaries. - [Security & Control](https://monk.io/security): Understand permissions, approvals, credential handling and runtime controls. - [Manifesto](https://monk.io/manifesto): Read why Monk is building the Universal Operator. ## Why add Monk to a coding agent? Coding agents can already deploy through commands and tools. Monk maintains a live graph of the services, infrastructure and connections it manages. A fresh session can read deployment state and current status, then prepare changes for approval. Permissions, credential handling and orchestration supply the controls and execution around that graph. This reduces repeated discovery; no measured token saving is claimed. Cost estimates and tracking depend on supported resources and available pricing data. The homepage illustration and product walkthrough use synthetic state and costs. ## Compatibility Coding-agent hosts: Claude Code, Codex, Cursor. Cloud infrastructure providers: AWS, Google Cloud, Azure, DigitalOcean, Hetzner. Managed-service integrations include Netlify, Vercel, Cloudflare, MongoDB Atlas, Redis Cloud, Neon. These are separate from infrastructure providers; availability depends on the integration, region and permissions. The catalog is a starting point, not a closed list. Monk runs container workloads and supports custom API integrations. Custom integrations need implementation and testing; provider APIs, authentication, permissions and runtime requirements still apply. The selected provider and integration determine permissions, quotas, service features, and regional availability. Code analysis targets deployment signals such as entry points, dependencies, environment variables, endpoints, database connections, and existing configuration. It does not read every source line or recreate unknown infrastructure state. Existing deployments require separate planning for data transfer, secrets outside the source tree, traffic cutover, and any provider-specific migration work. ## Product - [Integrations](https://monk.io/integrations): A searchable directory of cloud accounts, managed services, runnable packages and coding-agent connections. Available actions depend on the package and your provider permissions. - [Monk — The Universal Operator](https://monk.io/): Monk is the operator for the software your agents write. Your coding agent writes it; Monk ships it and keeps it running — on your cloud, on your bill. - [Monk for ops and platform teams](https://monk.io/for/ops): Team evaluation: permissions, approvals, recovery, cloud ownership and coexistence. ## Get started - [Deploy alongside your app.](https://monk.io/start/existing): Monk builds fresh infrastructure from your codebase. Your existing deployment stays untouched. Test the new environment before you switch. - [Put your new app to work.](https://monk.io/start/new): Set up the plugin, analyze your repository and review the deployment plan. ## For coding agents - [Monk for coding agents — setup and operating boundaries](https://monk.io/agents): Plugin setup, first requests, prerequisites and operating boundaries. ## Use cases - [Use cases](https://monk.io/use-cases): GPU operations, Hetzner economics, application deployment and branch environments. - [GPU pods that start where the stock is — RunPod and Monk](https://monk.io/use-cases/healthtech-gpu-runpod): Pods start where the GPUs are, and stop when the work does. - [Run your app on Hetzner with Monk](https://monk.io/use-cases/hetzner-managed-feel): Hetzner prices, with Postgres, Redis and the queue run for you. - [Deploy your app in your cloud](https://monk.io/use-cases/ship-what-your-agent-wrote): From a repository to an app with its database, storage and service connections. - [Operate your app with a small team](https://monk.io/use-cases/production-without-a-platform-team): Inspect the system, recover workloads and review fixes without building a platform team first. - [Separate branch environments with Capsules](https://monk.io/use-cases/capsules-for-parallel-agents): Test branches with their dependencies, without competing for one shared staging environment. ## Comparisons - [Compare deployment and operations approaches](https://monk.io/compare): Compare operating responsibilities, state, controls, adoption and costs. - [Monk vs your coding agent alone — deployment and operations](https://monk.io/compare/diy-with-your-coding-agent): A persistent application model and controlled operations, beyond one conversation. - [Monk vs Terraform — deployment and operations](https://monk.io/compare/terraform): Infrastructure as code and the work of running the application. - [Monk vs Kubernetes — deployment and operations](https://monk.io/compare/kubernetes): The runtime platform and the services around it. - [Monk vs a deployment platform — deployment and operations](https://monk.io/compare/paas): Hosted infrastructure or an application in your own cloud account. - [Monk vs Railway — deployment and operations](https://monk.io/compare/railway): Railway supports agent workflows and whole-environment PR previews. Those capabilities are not exclusive to Monk. - [Monk vs Render — deployment and operations](https://monk.io/compare/render): Render Blueprints define interconnected services, databases and environment groups in YAML. Blueprint sync can apply changes and recreate missing resources. - [Monk vs Heroku — deployment and operations](https://monk.io/compare/heroku): Heroku Review Apps create temporary apps for pull requests, with setup configured through app.json and the pipeline workflow. - [Monk vs Fly.io — deployment and operations](https://monk.io/compare/fly): Fly.io exposes app configuration for Machines, processes, services, health checks and scaling behavior. It gives developers direct control over many runtime choices. ## Authoritative references - [Plugin installation](https://github.com/monk-io/monk-plugin): Complete installation and authentication for Claude Code, Codex and Cursor. - [First deployment](https://docs.monk.io/getting-started/first-deployment): Prerequisites, cloud credentials and the deployment workflow. - [Watcher](https://docs.monk.io/features/watcher): AI diagnosis and remediation approval. - [Pricing](https://monk.io/pricing): Capacity plans, managed components and separate cloud costs. - [Security](https://monk.io/security): Current security practices and program. - [Documentation](https://docs.monk.io/): Supported features, integrations and operation details.