Agent Mesh
Go a2aa2a-protocolagent-meshagentic-aiinferencemcpmeshmesh-networknetwork-securitynetworking-aip2p-network
# Bare-Metal GPU Reasoning Sidecar Node (node-inference.yaml).
#
# Sits beside the self-hosted open-weights LLM server (Ollama / vLLM serving
# DeepSeek-R1 or Gemma 3) and publishes it to the mesh as inference://private-llm.
# Enforces an inbound Datalog check so only on-premises private nodes may
# consume GPU inference capacity.
version: "v1alpha1"
labels:
residency: on-prem
zone: gpu-pool
dept: finance
attenuation:
checks:
- 'check if label("residency", "on-prem");'
services:
- type: inference
name: private-llm
description: "Self-hosted DeepSeek-R1 / Gemma 3 reasoning engine on bare-metal GPU"
target_url: "http://127.0.0.1:18892"