Skip to Content
DocsOperator Overview

Operating Agor

These guides are for people responsible for an Agor deployment or workspace administration. If your workspace is already running, start with Using Agor and Sharing & Permissions instead.

Install and choose an execution model

  • Installation options: packaged, Docker, and source installations.
  • Deployment configuration: daemon settings, authentication configuration, telemetry, and configuration ownership.
  • Execution isolation: choose trusted local simple, fail-closed local sandbox, or explicitly configured delegated execution.
  • Containerized execution: provision an external executor substrate and persistent workspaces. A launcher template alone does not provide isolation.

Running executors in containers and running multiple daemon replicas are separate choices. For multiple daemons, follow Daemon High Availability: active-active mode requires PostgreSQL, Redis, sticky Engine.IO ingress, and a supported storage topology. SQLite on shared storage is not a substitute.

Secure access and connections

Deployment-global settings belong to the deployment owner. Tenant administrators cannot change those settings through workspace preferences. MCP security modes are tenant-owned API settings, not config.yaml keys; personal connections and credentials remain user-owned.

Maintain and upgrade

Before an upgrade, check the migration instructions relevant to your deployment. Some changes require stopping all daemons and taking tested backups; they are not rolling upgrades.

For repository contributions and local test environments, use the Development Guide, not a production deployment runbook.

Last updated on