How xinity-ai/xinity-ai works
Xinity is a self-hosted enterprise AI operations platform: users manage organizations, access, deployments, and recorded usage in the dashboard; API clients send OpenAI-compatible inference requests through the gateway to models running on daemon-managed nodes. The README documents the gateway-to-daemon tether and Ollama/vLLM engines. The tree and sampled code also support API/MCP, usage data, and model-catalog services. Some subsystem links reflect documented capabilities rather than inspected implementation; uncertain internal wiring is omitted.
Main entry points: +page.server.ts Management dashboard · index.ts Node daemon · openai.ts OpenAI gateway · router.ts Management API
Components 22
Control plane
- Management dashboard
packages/xinity-ai-dashboard/src/routes/(authenticated)/+page.server.ts - Authentication and SSO
packages/xinity-ai-dashboard/src/lib/server/auth-server.ts - Management API
packages/xinity-ai-dashboard/src/lib/server/orpc/router.ts - MCP endpoint
packages/xinity-ai-dashboard/src/routes/mcp/+server.ts - Deployment orchestration
packages/xinity-ai-dashboard/src/lib/server/lib/orchestration.mod.ts - Compute management
packages/xinity-ai-dashboard/src/lib/server/compute/compute.ts
Inference plane
- OpenAI gateway
packages/xinity-ai-daemon/src/modules/serverfront/openai.ts - Gateway proxy
packages/xinity-ai-dashboard/src/lib/server/gateway-proxy.ts - Node daemon
packages/xinity-ai-daemon/src/index.ts - Tether bridge
packages/xinity-ai-daemon/src/modules/tether-client.ts - Model installation
packages/xinity-ai-daemon/src/modules/model-installation/ollama.ts - Model registry
packages/xinity-ai-daemon/src/modules/model-registry.ts
Platform state
- PostgreSQL
packages/common-db/src/connection.ts - Usage and call data
packages/common-db/src/schema/call-data.ts - Audit and governance
packages/xinity-ai-dashboard/src/lib/server/audit/audit-sink.ts - Metrics and observability
packages/xinity-ai-dashboard/src/lib/server/metrics.ts
External interfaces
- Operations CLI
packages/xinity-cli/src/lib/act/routes.ts - OpenAPI description
packages/xinity-ai-dashboard/src/routes/api/openapi.json/+server.ts
Model information
- Model catalog service
packages/xinity-infoserver/server.ts
Other components
- Operator
- API client
- Inference engines
Connections 25
- Operator → Management dashboard: manages
- API client → OpenAI gateway: requests inference
- Management dashboard → Authentication and SSO: authenticates
- Management dashboard → Management API: calls management API
- MCP endpoint → Management API: dispatches tools
- Management dashboard → PostgreSQL: reads and writes
- Management dashboard → Usage and call data: queries and rolls up
- Management dashboard → Deployment orchestration: manages deployments
- Deployment orchestration → PostgreSQL: persists configuration
- Management dashboard → Gateway proxy: forwards inference
- Gateway proxy → OpenAI gateway: proxies requests
- OpenAI gateway → Tether bridge: routes to nodes
- Tether bridge → Node daemon: bridges node traffic
- Node daemon → Model installation: manages models
- Node daemon → Inference engines: runs inference
- Node daemon → Model registry: tracks models
- Management dashboard → Compute management: manages capacity
- Management dashboard → Audit and governance: records governance events
- Management dashboard → Metrics and observability: exposes observability
- API client → MCP endpoint: calls tools
- API client → Model catalog service: resolves model metadata
- Model catalog service → Model registry: serves model catalogs
- Operator → Operations CLI: operates platform
- Operations CLI → Management API: discovers routes
- Operator → OpenAPI description: browses API schema