Systems Manager
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| HOST | No | Host to bind for HTTP transport | 0.0.0.0 |
| PORT | No | Port to bind for HTTP transport | 8000 |
| TRANSPORT | No | Transport mode (stdio or streamable-http) | stdio |
| SYSTEMS_API_KEY | Yes | The API key for Systems Manager |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
Systems Manager MCP Server
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
| logging | {} |
| prompts | {
"listChanged": true
} |
| resources | {
"subscribe": false,
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| health_checkD | – |
| get_management_capabilitiesD | – |
| get_process_detailsB | Retrieves deep cross-platform process details (threads, modules, memory). CONCEPT:SM-OS.deployment.deep-introspection-telemetry: Deep Introspection Telemetry |
| get_network_connectionsB | Maps active TCP/UDP endpoints directly to owning processes. CONCEPT:SM-OS.deployment.deep-introspection-telemetry: Deep Introspection Telemetry |
| capture_system_snapshotB | Takes a point-in-time snapshot of the system state (CPU, RAM, Processes). CONCEPT:SM-OS.deployment.deep-introspection-telemetry: Deep Introspection Telemetry |
| list_servicesB | Cross-platform service enumeration (systemctl or Get-Service). CONCEPT:SM-OS.deployment.package-service-mutation: Package & Service Mutation |
| manage_serviceC | Start/Stop/Restart/Enable/Disable services cross-platform. CONCEPT:SM-OS.deployment.package-service-mutation: Package & Service Mutation |
| list_kernel_modulesA | List loaded drivers/modules (lsmod or driverquery). CONCEPT:SM-OS.deployment.deep-introspection-telemetry: Deep Introspection Telemetry |
| query_system_logsB | Cross-platform log querying (journalctl or Get-WinEvent). CONCEPT:SM-OS.deployment.deep-introspection-telemetry: Deep Introspection Telemetry |
| start_system_traceC | Start a kernel-level event trace (ETW on Windows, or strace on Linux). |
| stop_system_traceC | Stop a kernel-level event trace. |
| sm_storage_health_smartB | Smart (sm_storage_health). Physical disk + BMC drive-fault health: SMART (incl. megaraid passthrough), BMC/IPMI drive-slot faults, and RAID physical-disk state — correlated so a BMC-flagged disk with clean SMART media reads as a link/aging fault, not media wear (CONCEPT:SM-OS.governance.sys-8/SYS-1.5). |
| sm_storage_health_faultsB | Faults (sm_storage_health). Physical disk + BMC drive-fault health: SMART (incl. megaraid passthrough), BMC/IPMI drive-slot faults, and RAID physical-disk state — correlated so a BMC-flagged disk with clean SMART media reads as a link/aging fault, not media wear (CONCEPT:SM-OS.governance.sys-8/SYS-1.5). |
| sm_storage_health_raidB | Raid (sm_storage_health). Physical disk + BMC drive-fault health: SMART (incl. megaraid passthrough), BMC/IPMI drive-slot faults, and RAID physical-disk state — correlated so a BMC-flagged disk with clean SMART media reads as a link/aging fault, not media wear (CONCEPT:SM-OS.governance.sys-8/SYS-1.5). |
| sm_storage_health_reportB | Report (sm_storage_health). Physical disk + BMC drive-fault health: SMART (incl. megaraid passthrough), BMC/IPMI drive-slot faults, and RAID physical-disk state — correlated so a BMC-flagged disk with clean SMART media reads as a link/aging fault, not media wear (CONCEPT:SM-OS.governance.sys-8/SYS-1.5). |
| sm_system_operationsC | System operations for managing packages, system health, and updates |
| sm_service_operationsC | Operations for managing system services |
| sm_process_operationsC | Operations for managing system processes |
| sm_network_operationsC | Operations for network analysis |
| sm_disk_operationsC | Operations for managing system disks |
| sm_user_operationsC | Operations for user and group management |
| sm_file_operationsC | Operations for file and log management |
| sm_cron_operationsC | Operations for cron jobs |
| sm_firewall_operationsC | Operations for firewall management |
| sm_advanced_operationsD | Operations for SSH and specialized managers |
| systems_ingest_hostC | Ingest a keyed host telemetry projection through the governed ChangeEnvelope boundary as typed HardwareNode, NetworkInterface, and DiskVolume nodes. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| connector-status | Ask for this connector server's current operational status. |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| _connector_status_resource | Native resource exposing this server's static status payload. |
| ontology://systems-manager-mcp/systems.ttl | |
| connector-resources | Data sources, ontology mappings and resources of this connector. |
| nic-bond-provisioner/SKILL.md | Convert multi-NIC Linux hosts to a systemd-networkd active-backup bond on a static canonical IP, retiring NetworkManager, with a console-safe dead-man auto-revert. Use when a node's egress/data-path is non-deterministic across reboots (multiple DHCP NICs), when stabilizing a Docker Swarm VXLAN data-path before a swarm re-init, or when the user asks to bond NICs, set a static canonical IP via netplan, or fix cross-node overlay drift. Do NOT use for single-NIC hosts (already deterministic), for LACP/802.3ad throughput bonding (this is failover, not aggregation), or for non-netplan distros. |
| nic-bond-provisioner/_manifest | File listing for nic-bond-provisioner |
| systems-manager-development/SKILL.md | Develop, test, and release the systems-manager Python connector. Covers the cross-platform systems manager, governed FastMCP server, connector source declarations, packaged skills and ontology, documentation, and the repository's shared delivery gates. Use before changing a provider tool, MCP boundary, connector content, or release metadata. |
| systems-manager-development/_manifest | File listing for systems-manager-development |
| systems-manager-operations/SKILL.md | Operate hosts and Agent OS services through the governed systems-manager MCP provider and Arbora delegation. Use for OS and package lifecycle, services, processes, network and disk inspection, storage or BMC health, firewall and managed-file work, agent identity, specialist registry, scheduler, watchdog, host telemetry ingestion, troubleshooting, and evidence-backed verification. |
| systems-manager-operations/_manifest | File listing for systems-manager-operations |
| ipmi-bmc-manager/SKILL.md | IPMI / BMC management atomic skill. Controls server power, opens Serial-over-LAN consoles, reads sensors and event logs, and configures the BMC LAN on out-of-band controllers (Dell iDRAC, HPE iLO, Supermicro, generic IPMI 2.0) via ipmitool — in-band or over LAN. Use for remote power control and for recovering a node whose OS network is unreachable. |
| ipmi-bmc-manager/_manifest | File listing for ipmi-bmc-manager |
TDQS
Scored across 26 tools
The set mixes specific tools (manage_service, list_services, get_process_details, get_network_connections) with broad catch-all buckets (sm_service_operations, sm_process_operations, sm_network_operations) that clearly overlap them, so an agent cannot tell which to pick for e.g. restarting a service or inspecting a process. The four sm_storage_health_* tools carry identical descriptions and differ only by a suffix (smart/faults/raid/report), making them essentially indistinguishable without guessing.
All names are snake_case, but conventions diverge sharply: bare verb_noun tools (list_services, query_system_logs), an sm_ prefixed generic family with a vague '_operations' tail, sm_storage_health_<noun> with domain-first ordering, and a lone systems_ingest_host. Suffix-only differentiation within the storage family and undescribed names like health_check add to the inconsistency.
At 26 tools the surface is heavy for a systems manager, and the excess is largely redundant: roughly ten generic '_operations' tools duplicate more targeted tools elsewhere in the set. A tighter, deduplicated set would serve the domain better.
Coverage across services, storage/SMART, network, users, cron, firewall, files, processes, logging, tracing and telemetry ingest is broad, so major operation classes exist. However, two tools (health_check, get_management_capabilities) have no descriptions, and it is unclear whether the generic operation buckets fully implement create/update/delete lifecycles, leaving gaps hard to verify.