Mission Control
The control plane above your 10 products — wire up the outside world and the automation, then hand scoped access down to the modules.
Wire up external systems once — firewalls, cloud, EDR, SIEM, scanners, ticketing, IdP — and broker scoped, encrypted credentials to the modules by reference.
Schedule and run agentic automation on an in-process scheduler (no Temporal dependency, ENG-1098) — connector sync and Assurance “breathing” continuous-control-monitoring — each with a run history.
One place to control who-can-do-what: roles and per-module / per-connection / per-agent grants, enforced by shared middleware.
The consolidated home for cross-module settings — branding, model/LLM providers, email, data classification, framework defaults.
The audit feed — every connection, grant, role, agent and config change, with who/what/when.
The operational feed — AI-generation, connector, import and doc-gen errors across every product, with the upstream message, so failures are visible without pod/shell access (incl. on exports).
Every LLM call across the suite — model, tokens, latency, status (and, when opted in, prompt/response) — kept in this tenant's own database. Native, no-egress alternative to LangSmith (ADR-0019).