// the find
ascending-llc/jarvis-registry
Connect any AI copilot or autonomous agent to your enterprise tools — through a single, secure MCP/Agent gateway with built-in identity, access control, and full observability.
Jarvis Registry is an enterprise MCP gateway that sits in front of your AI copilots and agents, handling auth, access control, and observability in one place. It builds on top of agentic-community/mcp-gateway-registry and extends it with OAuth 2.0/OIDC support, A2A agent orchestration, and a proper UI. The target audience is platform/infrastructure teams trying to govern AI tool access across an organization.
The IAM story is genuinely thought through — supporting Keycloak, Cognito, and Entra ID out of the box means you're not writing custom auth glue for whichever IdP your enterprise already uses. The observability stack (OpenTelemetry + Prometheus + pre-built Grafana dashboards in the repo) is included and wired up, not left as an exercise. Semantic search over registered MCP servers so agents can discover tools at runtime is a real solution to a problem most people solve manually. The deployment story is practical — Docker Compose for local, Helm-style guides for EKS/AKS/GKE, and the structure supports both.
The repo is a fork/evolution of agentic-community/mcp-gateway-registry, and that ancestry is visible: the codebase spans Python services, a TypeScript CLI, shell scripts, and a React frontend without strong cohesion. The credentials-provider directory is a pile of ad-hoc scripts rather than a designed subsystem. Test coverage appears thin — the auth-server has integration tests, but the gateway core and registry API have nothing visible in the tree. The 'enterprise-grade' framing outpaces the maturity of the actual project; this is still pre-1.0 and the architecture docs are aspirational in places.