Source: https://method.chakril.site/docs/deployment-lifecycle

**Documentation** [Host an existing MCP](https://method.chakril.site/docs/host-existing-mcp) [Client compatibility](https://method.chakril.site/docs/client-compatibility) [Agent-assisted setup](https://method.chakril.site/docs/agent-onboarding) [Quickstart](https://method.chakril.site/docs/quickstart) [Authentication](https://method.chakril.site/docs/authentication) [Deployment lifecycle](https://method.chakril.site/docs/deployment-lifecycle)

Guide

# Deployment lifecycle

Operate revisions safely from queue to deletion.

## Create

A deployment starts queued, is claimed by a connected runtime cluster, and becomes ready only when its workload passes health checks.

## Update

Saving a reviewed definition creates an immutable version. An update keeps the last healthy runtime serving until the replacement is ready.

## Pause and resume

Pause stops serving the endpoint while retaining the deployment record and configuration. Resume schedules it again.

## Delete

Deletion requires the endpoint slug as confirmation. The reconciler removes the runtime before finalizing the database state.

Ready to put this into practice? [Open MCP Ops](https://method.chakril.site/login)