UNIVERSAL GATEWAY · FEATURES
Everything a production gateway needs — and the parts agents need too.
Routing, auth, transformations, multi-tenancy, logging, alerting, and a plugin system, built to handle multi-protocol agent traffic as a first-class citizen rather than an afterthought.
Core capabilities.
Multi-protocol routing & circuit breaker
Route by host, path, header, or protocol to any upstream. Per-route circuit breakers trip on error-rate or latency thresholds so one failing backend never takes the rest down.
Auth, OAuth2 & transformations
API keys and OAuth2 at the edge, plus request/response transformations — rewrite headers, bodies, and paths — and API versioning so clients migrate on their own schedule.
Multi-tenant quotas & usage
Isolate tenants with their own keys, rate limits, and quotas. Per-tenant usage counters feed billing, capacity planning, and abuse detection.
Structured logging & export
JSON logs streamed to Elasticsearch, Splunk, Amazon S3, CloudWatch, or Datadog. Every request carries a correlation ID across protocols and upstreams.
Email & Slack alerts
Threshold and anomaly alerts delivered to email and Slack, so on-call hears about a tripped breaker or a quota breach before customers do.
Plugin system & CLI
Extend routing, auth, and transforms with first-party and custom plugins. Drive everything from the `agentanywhere-gateway` CLI for scripting and CI.
Observability: RED, SLOs, tracing
Rate, Errors, Duration metrics in Prometheus format, SLO tracking with burn-rate alerts, and distributed tracing exported over OTLP.
Governance: audit, RBAC, policies
A tamper-evident audit chain, role-based access control, declarative policies, and IP allow/deny rules — the controls a regulated deployment is required to show.
Production-ready by default.
The gateway ships with health and readiness endpoints, graceful shutdown, configurable timeouts and retries, and a Prometheus `/metrics` endpoint — the things you would otherwise bolt on before a launch.
It runs as a single container or scales horizontally behind your load balancer. Configuration is environment-driven, so the same image promotes cleanly from local to staging to a sovereign production deployment.
See the full capability surface.
The documentation covers every route option, plugin hook, transformation, and export target, with copy-paste configuration. Start there, or talk to us about a governed deployment.