Verified

Nomad MCP Server

HashiCorp Nomad cluster interaction via 110 tools for observability and management.

Data last scanned today · Reviewed today

40 32 dangerous tools110 tools★ 4GoMPL-2.0

Overview

Nomad MCP Server provides AI models with structured access to HashiCorp Nomad clusters. It allows assistants to inspect jobs, allocations, nodes, and deployments, enabling them to troubleshoot placement issues, analyze task logs, and query cluster capacity.

Our verdict

This server carries a high risk profile with a security score of 30/100. Because it is community-maintained rather than official and lacks an independent editorial review, users should exercise caution. Its security relies heavily on your configuration of Nomad ACL tokens and environment variables like NOMAD_MCP_READ_ONLY. It is suitable for development environments, but production usage requires a strict, read-only token policy scoped to specific namespaces.

Supports over 100 tools for cluster management and diagnosis
Zero direct dependencies minimize the software supply chain surface
Configuration follows patterns used by HashiCorp's official Vault MCP
Built-in opt-in controls for read-only and namespace restrictions

Community-maintained project lacking official HashiCorp support
Authentication methods and security posture are not yet audited
Low contributor count suggests a single-point-of-failure maintenance risk

Tools

Tool	Description	Risk
find_problems	One ranked list of everything currently wrong: failed and lost allocations, blocked evaluations, stuck deployments, queued work, unhealthy nodes.	low
search_job_logs	Grep every allocation of a job at once, concurrently, returning only matching lines with the allocation, node and task they came from.	low
build_job_timeline	Merge job versions, evaluations, deployments and task events into one chronological narrative.	low
diagnose_volume	Follow a volume through its CSI plugin, its claims and the nodes holding them.	low
diagnose_integrations	Find Vault and Consul failures from Nomad's own task events.	low
get_cluster_capacity	What the cluster has, what is allocated, and the largest task group placeable on any single node.	low
explain_placement	Node by node, whether a job's groups fit.	low
analyze_job_resources	What each task reserved against what it is observed using, plus OOM-kill history.	low
check_connection	Address, TLS, token, ACL state, edition and permission probes.	low
get_cluster_status	Leader, server peers and versions, edition, node counts by state.	low

Compatibility

Client	Local	Docker	Remote	Read-only
ChatGPT
Claude Desktop
Cursor
VS Code
Windsurf

Frequently asked questions

›How do I restrict the server to only read information?

By default, the server runs with NOMAD_MCP_READ_ONLY=true, which refuses any tool that attempts to change the cluster state.

›Can I prevent the AI from accessing Nomad Variables?

Yes, by setting NOMAD_MCP_ALLOW_VARIABLE_READS to false, the read_variable tool will not return values, though list_variables will still return path names.

›What is the best way to secure access to my Nomad cluster?

The most effective control is providing a Nomad ACL token scoped only to the specific namespaces and permissions required, rather than using a full management token.

›How do I troubleshoot connection issues with the server?

You can ask the AI model to run the check_connection tool, which identifies which part of the connection chain is failing and provides guidance on how to fix it.

›Does this server support both Nomad Community Edition and Enterprise?

Yes, the server works against both Nomad Community Edition and Enterprise, regardless of whether the cluster is running locally, on EC2, in Docker, or elsewhere.