Verified

Janus MCP Server

Kubernetes cluster operations with automated secret redaction for AI assistants.

Data last scanned today · Reviewed today

384 tools★ 0PythonApache-2.0

Overview

Janus acts as a local intermediary between your LLM client and your Kubernetes API server. By housing your KUBECONFIG locally, it ensures that sensitive authentication data never leaves your environment, providing AI agents with restricted access to your cluster through specific, tool-based interactions instead of raw data dumps. The server emphasizes safety through automatic output sanitization, stripping secrets, tokens, and sensitive environment variables from API responses. It further enforces control by requiring explicit human approval for destructive operations like scaling or deleting resources, allowing models to propose changes while keeping the final execution in user hands.

Our verdict

With a security score of 33/100, Janus is categorized as high risk. This rating reflects its status as a community-maintained project that has not yet undergone professional editorial or security reviews. While it features a low-surface-area design with zero direct dependencies, users should exercise caution as the authentication and read-only implementation details lack verification. It is best suited for developers willing to perform their own due diligence in restricted environments.

Automatic redaction of secrets and sensitive metadata
Requires human confirmation for destructive cluster actions
Zero direct dependencies minimize supply-chain risks
Limits exposure by keeping KUBECONFIG local

High risk score due to lack of security review
Community-maintained with no official vendor backing
Authentication methods have not been audited

Tools

Tool	Description	Risk
get_pods	Fetches pods from the cluster.	low
describe_deployment	Describes a specific deployment.	low
get_events	Retrieves cluster events.	low
get_cluster_summary	Provides a summary of the cluster status.	low

Compatibility

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

Frequently asked questions

›How does Janus protect sensitive Kubernetes data?

Janus automatically redacts secrets, tokens, environment variable values, and sensitive metadata from Kubernetes API responses before they are sent to the LLM.

›Can an LLM perform destructive actions on my cluster?

While the LLM can propose actions like scale or delete, any destructive operation requires explicit human approval within your MCP client before it is executed.

›Does Janus require me to share my KUBECONFIG with the AI?

No, your KUBECONFIG remains local to the machine or process running the Janus server and is never transmitted to the LLM.

›Can I limit the server to specific clusters or namespaces?

Yes, Janus allows you to configure scoped access by locking it to specific namespaces, clusters, or subsets of resources in your configuration file.

›What methods are available to install the Janus server?

You can install the server via PyPI using pipx or uv, as a Homebrew package, or as a container image from GitHub Container Registry.