The profile switch is what sets it apart: anyone running a production and a staging tenancy spends real time restarting tools to change which one they are pointed at. Here it is a call, and a follow-up call tells you which tenancy you are currently talking to — worth making before anything that changes state, because the same connection can stop a DB node or reboot a cluster.
A read-and-manage layer over Oracle Cloud Infrastructure that authenticates through your existing OCI CLI config. Profiles can be switched at runtime, so one session can cross tenancies without a restart, and the surface reaches well past compute — identity, networking, storage, databases including Exadata, Kubernetes, cost and monitoring.
- Profiles handled live: list_oci_profiles reads what is in your OCI config, set_oci_profile activates one, and get_current_oci_profile says which tenancy you are actually talking to.
- Identity: list_compartments, plus users, groups, policies and dynamic groups listed and read in detail — get_user returns group memberships, get_policy returns the statements themselves.
- Compute: list_instances and get_instance, with start_instance and stop_instance for lifecycle, and list_images, list_shapes, list_availability_domains and list_fault_domains for what you can launch and where.
- Databases at every level: DB Systems and DB Nodes with start, stop, reboot and reset; list_databases and list_autonomous_databases for the databases themselves; Data Guard associations with apply lag and protection mode; list_backups and maintenance runs.
- Exadata Cloud Service: cloud Exadata infrastructures, VM clusters with their RAC nodes, SCAN DNS and versions, and autonomous VM clusters.
- Networking: VCNs and subnets, VNICs, security lists with their ingress and egress rules, and network security groups.
- Storage: Object Storage namespace and buckets, block volumes and boot volumes, and File Storage file systems.
- Load balancers, both classic and network, with their backend sets and listeners.
- Kubernetes: list_oke_clusters and get_oke_cluster with add-ons and available upgrades, node pools down to individual nodes, work requests for async operations, and get_oke_cluster_kubeconfig for kubectl access.
- Cost and monitoring: cost summaries by service or compartment, budgets with actual and forecast spend, alarms with their history, list_metrics and query_metric_data over MQL, and search_logs across log groups.
- KMS vaults and encryption keys, with algorithm, shape and versions.
A configured OCI CLI — the server reads its profiles and credentials from your OCI config — and permissions in the tenancy for whatever you ask it to do. Python 3.10 or newer. You can start it without naming a profile and choose one in conversation, or preload one at launch.