jaimenbell/vllm-ops-mcp

Overview

A read-only MCP server that exposes operational health data for a vLLM instance running under WSL2/systemd. It reports liveness versus real-completion health tiers, GPU/VRAM usage, systemd service state, and live serve-flag introspection.

Best for

Best for
Developers running vLLM on WSL2/systemd who need a lightweight, read-only health monitoring MCP server.

Use cases

• Monitor vLLM instance health from an MCP-compatible client
• Check GPU/VRAM utilization and systemd service status
• Introspect live serve flags to verify runtime configuration

Notes

A read-only MCP server that exposes operational health data for a vLLM instance running under WSL2/systemd. It reports liveness versus real-completion health tiers, GPU/VRAM usage, systemd service state, and live serve-flag introspection.

0 stars on GitHub. Last updated 2026-08-12. Licensed MIT.

Use cases

• Monitor vLLM instance health from an MCP-compatible client
• Check GPU/VRAM utilization and systemd service status
• Introspect live serve flags to verify runtime configuration

Pros

• Read-only design reduces risk of accidental state changes
• Distinguishes liveness from actual completion readiness
• Covers both system-level and model-serving health signals

Cons

• Limited to WSL2/systemd environments
• No write or control operations, only monitoring
• Zero stars suggests minimal community validation

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.