Alpha Plugin
nexus-llm-bridge
v0.1.0AlphaConnect your own vLLM/LLM inference stack and auto-configure it as a dedicated provider in OpenCode workspaces.
Overview
Bring your own LLM hardware into the Nexus ecosystem. Register any vLLM-compatible inference endpoint (local GPU server, cloud VM, or cluster) and Nexus auto-discovers available models, monitors health, and writes the provider configuration directly into your opencode.json. Includes failover to cloud providers when local hardware is unavailable.
Key Features
- Auto-discovery of vLLM / LiteLLM endpoints
- Health monitoring + failover to cloud
- Auto-write to opencode.json providers block
- Per-project model assignment via UI
Registered Hooks
Events this plugin listens to in the OpenCode runtime.
provider.healthPeriodically checks registered LLM endpoints for availability and model readiness.
Custom Tools
MCP tools this plugin exposes to agents.
nexus_llm_registerRegister a new vLLM endpoint with connection details and model discovery.
nexus_llm_statusReports health, latency, and throughput statistics for all registered LLM endpoints.