The full upstream README, mirrored here for reference. Install config, tool schemas, adoption signals, and an original overview live on the ProxmoxMCP Plus listing page.
Operate Proxmox VE from MCP clients, AI agents, and OpenAPI tooling through one security-conscious control plane for VMs, LXCs, snapshots, backups, ISOs, container commands, and persistent long-running jobs.
Quick Start | Client Install | Demo | Tools | Safety | Scenarios | Docs | Wiki
ProxmoxMCP-Plus sits between AI clients and Proxmox VE so operators do not have to stitch together raw API calls, one-off shell scripts, and custom job polling for every workflow.
It exposes the same operational surface in two ways:
MCP for Claude Desktop, Cursor, VS Code, Open WebUI, Codex, and other MCP-capable agentsOpenAPI for HTTP automation, dashboards, internal tools, and no-code workflowsWhat you get:
| Priority | How the project handles it |
|---|---|
| Dual access paths | Native MCP for agent workflows and OpenAPI for standard HTTP automation |
| Proxmox-oriented workflows | Day-2 VM, LXC, snapshot, backup, ISO, storage, and cluster operations |
| Long-running operations | Stable job_ids, Proxmox UPID tracking, polling, retry, cancel, and audit history |
| Safer execution | Proxmox API tokens, OpenAPI bearer auth, command policy, approval tokens, TLS validation, and MCP HTTP Host/Origin controls |
| Real validation | Unit, integration, Docker/OpenAPI, and live Proxmox e2e entry points are documented in the repo |
Create a Proxmox API token with only the permissions your workflows need. Then create the local config file:
Then edit proxmox-config/config.json with your environment. At minimum, it needs:
proxmox.hostproxmox.portauth.userauth.token_nameauth.token_valueAdd an ssh section as well if you want container command execution.
Add a jobs section if you want job state persisted somewhere other than the default local SQLite file.
For real live verification, use a separate proxmox-config/config.live.json created from proxmox-config/config.live.example.json.
Do not point live e2e at a placeholder or local-only config.json unless you intentionally run a local API tunnel there.
Optional job persistence config:
Optional tool exposure filtering can reduce the schemas sent to MCP clients. It is
disabled by default, so existing configurations continue to expose every available
tool. Configure exactly one mode under mcp:
Alternatively, use tool_denylist, or the comma-separated environment variables
MCP_TOOL_ALLOWLIST and MCP_TOOL_DENYLIST. Do not configure both modes. Environment
selection replaces the file-level filtering mode. An empty allowlist exposes no tools;
an empty denylist hides none. Exact lowercase tool names are required, and unknown names
fail startup so a typo cannot silently widen access. Restart or reconnect the MCP server
after changing the filter.
| Path | Best for | Start command | Verify |
|---|---|---|---|
| MCP stdio from PyPI | Claude Desktop, Cursor, VS Code, Codex, local agents | uvx proxmox-mcp-plus | client lists get_nodes, get_vms, and job tools |
| Native MCP HTTP from Docker | remote MCP clients that support Streamable HTTP | docker compose --profile mcp-http up -d proxmox-mcp-http | connect to http://localhost:8000/mcp |
| OpenAPI bridge from Docker | HTTP clients, dashboards, scripts, no-code tools | docker compose up -d | curl -f http://localhost:8811/livez |
Or install it first:
Use this path when the MCP client launches a local stdio server.
Code Mode is disabled by default to preserve the legacy full tool catalog. Enable it
with mcp.code_mode: true in the config file or MCP_CODE_MODE=true. When enabled,
MCP exposes three tools instead: proxmox_code_search, proxmox_code_get_schema, and
proxmox_code_execute. Code execution runs in an isolated sandbox and reaches domain
tools through the existing validation, policy, and approval path. Discovery uses the
filtered runtime catalog, so it also works in installed wheels. For example:
The final expression is returned as data.result; tool results use MCP JSON content
blocks (and structured content where supplied). Use proxmox_code_get_schema for
arguments, including target and approval tokens. Scripts have no filesystem or
network access except registered tool calls. Limits: 64,000 source characters,
100 MB sandbox memory, 25 tool calls, 16 KB final JSON, and two concurrent executions.
Execution has a 30-second budget; cancelling or failing a script does not roll back
tool side effects. Do not automatically retry a failed mutation script.
Use this path when a remote MCP client supports Streamable HTTP:
Point MCP clients at:
Send Authorization: Bearer <MCP_API_KEY> with every native MCP HTTP request.
The key is independent of Proxmox credentials. Native Streamable HTTP and SSE
require a key by default. For an endpoint protected by an external access-control
layer such as Tailscale ACLs, explicitly set MCP_ALLOW_UNAUTHENTICATED_HTTP=true
or mcp.allow_unauthenticated_http: true in JSON configuration. This allows
keyless startup and logs a warning; it does not configure or verify the external
access controls. A configured MCP_API_KEY is always enforced, even with opt-out.
This setting does not change OpenAPI authentication or DNS rebinding protection.
When serving MCP HTTP behind a reverse proxy, keep DNS rebinding protection enabled and allow only the hostnames you expect:
OpenAPI mode is the default Docker runtime and requires an API key:
Verify the OpenAPI surface:
For local unauthenticated development only, set PROXMOX_ALLOW_NO_AUTH=true.
The 8811 service is the OpenAPI/REST bridge. The 8000 service is the native MCP HTTP endpoint.
Use the one-click buttons when your client supports MCP install deeplinks, or copy the JSON config below.
Recommended stdio config:
Use a local config file if you prefer not to keep credentials in the client config:
Client-specific examples for Claude Desktop, Cursor, VS Code, Codex, OpenCode, Open WebUI, Streamable HTTP, and OpenAPI are in the Client Setup Guide and Integrations Guide.
This demo is a direct terminal recording of qwen/qwen3.6-plus driving a live MCP session in English against a local Proxmox lab. It shows natural-language control flowing through MCP tools to create and start an LXC, execute a container command, and confirm the authenticated HTTP /health surface.

Start with read-only discovery, then move to mutating tools only after the target node, storage, VMID, and permissions are clear.
| Operator goal | Start with | Then use | Notes |
|---|---|---|---|
| Inspect the cluster | get_nodes, get_cluster_status | get_storage, get_vms, get_containers | Best first health check after client install |
| Create or manage a VM | get_nodes, get_storage | create_vm, start_vm, stop_vm, delete_vm | Long-running mutations return job_id and Proxmox task_id |
| Manage LXCs | get_containers, get_storage | create_container, start_container, stop_container, delete_container | SSH-backed command tools require the optional ssh config |
| Roll back risky changes | list_snapshots with vm_type=qemu or vm_type=lxc | create_snapshot, rollback_snapshot, delete_snapshot | Create a snapshot before destructive workflow tests |
| Run commands inside guests | VM or container status tools | execute_vm_command, execute_container_command | VM path needs QEMU Guest Agent; LXC path needs SSH to the Proxmox node |
| Track async work | mutation response with job_id | poll_job, get_job, list_jobs, retry_job, cancel_job | Use job_id for agent/user conversations and task_id for raw Proxmox traceability |
| Inspect logs | get_node_syslog, get_cluster_log | get_task_log, get_node_firewall_log, get_guest_firewall_log | All log tools are read-only; get_task_log accepts any Proxmox UPID |
| Automate from HTTP tools | /openapi.json | /jobs, /health, generated tool routes | Use bearer auth and keep CORS restricted outside local development |
For the full tool map, see the Tool Selection Guide and API & Tool Reference.
ProxmoxMCP-Plus is an access layer, not a replacement for Proxmox RBAC, network controls, or client-side MCP approval prompts.
The project gives operators several control points:
PROXMOX_API_KEY protects the OpenAPI bridge by default.MCP_API_KEY protects native Streamable HTTP and SSE with Bearer authentication; it is required unless MCP_ALLOW_UNAUTHENTICATED_HTTP=true explicitly delegates access control.command_policy controls command execution and high-risk operations.approval_token can gate command execution and high-risk mutating actions.Read the Security Guide before exposing the server outside a trusted local environment.
ProxmoxMCP-Plus provides a unified control surface for the operational tasks most teams actually need in Proxmox VE. The same server can expose these workflows to MCP clients for LLM and AI-agent use cases, and to HTTP consumers through the OpenAPI bridge.
Supported workflow areas:
| Capability Area | Availability |
|---|---|
| VM create / start / stop / delete | Available |
| VM snapshot create / rollback / delete | Available |
| Backup create / restore | Available |
| ISO download / delete | Available |
| LXC create / start / stop / delete | Available |
| Container SSH-backed command execution | Available |
| Container authorized_keys update | Available |
| Persistent job store for long tasks | Available |
MCP job control tools (list_jobs, get_job, poll_job, cancel_job, retry_job) | Available |
OpenAPI /jobs endpoints with explicit status codes | Available |
Local OpenAPI /livez, /readyz, /health, and schema | Available |
Docker native MCP Streamable HTTP at /mcp | Available |
Docker image build and /livez | Available |
Validation and contract entry points in this repository:
pytest -q --cov=proxmox_mcp --cov-report=term-missing --cov-fail-under=75ruff check .mypy src --ignore-missing-importspip-audit -r requirements.txttests/integration/test_real_contract.pytests/scripts/run_real_e2e.pytests/scripts/run_real_e2e.py now prefers proxmox-config/config.live.json or PROXMOX_MCP_E2E_CONFIG.
This avoids accidentally running live checks against a machine-specific default config.json.
Many Proxmox mutations are asynchronous. ProxmoxMCP-Plus now wraps those tasks in a persistent job layer so MCP and OpenAPI clients can track them through a stable Job ID.
Long-running tools such as VM create/start/stop, container create/start/stop, snapshot changes, backup/restore, and ISO download/delete now return both:
task_id: the raw Proxmox UPIDjob_id: the stable server-side job recordThe job record stores:
UPIDsBy default the job store persists to proxmox-jobs.sqlite3, so restart does not lose in-flight or completed job metadata.
list_jobsget_jobpoll_jobcancel_jobretry_jobWhen the OpenAPI proxy is enabled and a local JobStore is available, these routes are exposed directly:
| Path | Method | Purpose | Success Codes |
|---|---|---|---|
/jobs | GET | list persisted jobs | 200 |
/jobs/{job_id} | GET | fetch one job, optional refresh=true | 200 |
/jobs/{job_id}/poll | POST | refresh status from Proxmox | 200 |
/jobs/{job_id}/cancel | POST | request cancellation | 202 |
/jobs/{job_id}/retry | POST | replay a stored retry recipe | 202 |
Common error codes:
404: unknown job_id409: the job exists but that operation is not valid now503: the OpenAPI proxy was started without a local JobStoretests/scripts/run_real_e2e.py now prefers proxmox-config/config.live.json or PROXMOX_MCP_E2E_CONFIG.
This avoids accidentally running live checks against a machine-specific default config.json.
| Capability | Official Proxmox API | One-off scripts | ProxmoxMCP-Plus |
|---|---|---|---|
| MCP for LLM and AI agent workflows | No | No | Yes |
| OpenAPI surface for standard HTTP tooling | No | Usually no | Yes |
| VM and LXC operations in one interface | Low-level only | Depends | Yes |
| Snapshot, backup, and restore workflows | Low-level only | Depends | Yes |
| Persistent async job tracking and retry | No | Rare | Yes |
| Container command execution with policy controls | No | Custom only | Yes |
| Docker distribution path | No | Rare | Yes |
| Repository-level live-environment verification | N/A | Rare | Yes |
Ready-to-copy examples live in docs/examples/:
These are written for both human operators and LLM-driven usage.
The README is intentionally optimized for fast GitHub comprehension. Longer operational docs live in docs/wiki/ and can also be published to the GitHub Wiki.
| If you need to... | Start here |
|---|---|
| Understand the project and deployment flow | Wiki Home |
| Configure and run against a Proxmox environment | Operator Guide |
| Connect Claude Desktop, Cursor, VS Code, Codex, Open WebUI, or HTTP clients | Client Setup Guide |
| Choose the right tool for a workflow | Tool Selection Guide |
| Review docs quality goals, media plan, and publishing checklist | Documentation Quality Plan |
| Review integration patterns and transport details | Integrations Guide |
| Install from MCP-aware IDEs and agents | Agent Installation |
| Enable LXC command execution over SSH | Container Command Execution |
| Review security and command policy | Security Guide |
| Inspect tool parameters, prerequisites, and behavior | API & Tool Reference |
| Debug startup, auth, or health issues | Troubleshooting |
| Work on the codebase or release it | Developer Guide |
| Review release and upgrade notes | Release & Upgrade Notes |
Published wiki:
src/proxmox_mcp/: MCP server, config loading, security, OpenAPI bridgemain.py: MCP entrypoint for local and client-driven usagedocker-compose.yml: HTTP/OpenAPI runtimerequirements/: auxiliary dependency sources and runtime install listsscripts/: helper startup scripts for local workflowstests/scripts/run_real_e2e.py: live Proxmox and Docker/OpenAPI pathtests/: unit and integration coveragedocs/examples/: scenario-driven prompts and HTTP examplesdocs/wiki/: longer-form operator, integration, and reference docsParamiko 5.0.0 or newer is required so pip-audit can run without a CVE-2026-44405 exception.
Use list_isos to find an existing ISO volume (or download_iso to obtain one).
create_vm accepts iso_volume="local:iso/debian.iso", mounts it on ide3 by
default and boots CD-ROM before disk. update_vm_config can mount or change that
ISO, eject with iso_volume="none", set boot_order="scsi0;ide3", or change
network_bridge on net0 while retaining MAC/VLAN/firewall settings. Choose a
free cdrom_device; existing data disks and cloud-init drives are never replaced.
Media and bridge edits read current configuration and therefore need VM.Audit
as well as the relevant configuration privileges. Existing sizing/cloud-init-only
updates still do not require a preliminary read.
LXC uses OS templates, not installer ISOs. create_container retains DHCP by default
and now accepts network_bridge, ip="192.168.1.50/24", gw="192.168.1.1", ip6
and gw6. update_container_network edits those fields on an existing interface
(default net0, selectable through net31) and preserves all unspecified options.
Use an empty gateway string to remove it; changing to DHCP/manual removes the old
gateway for that address family. Network changes can interrupt guest connectivity.
Both edit tools follow named-target, read-only and high-risk approval policies.
Add update_container_network to custom high-risk lists and desired tool allowlists.
See v0.5.19 release notes for upgrade details.