Deployment topology
Every service, its dependencies, its ports, and its health contract.
Requirements & capacity planning
CPU, memory, disk, and GPU by service, sized against member count.
Install & upgrade
Validate, build, deploy, migrate, and roll back with the extension CLI.
Artifact renderer packs
Install, enable, verify, disable, and remove trusted offline Artifact renderers.
Configuration reference
Settings by domain, their defaults, and which ones carry secrets.
Feature flags & rollout
What each flag gates, the dependency order, and how to roll back safely.
Models & embeddings
Choosing chat, embedding, and reranker deployments, and what breaks if you choose wrong.
Connectors administration
Exposing discovered connector instances as tools, as RAG sources, or neither.
Tools & packages administration
Platform tools, packaged tools, and their approval boundaries.
Ingestion & indexing
How content becomes searchable, and how to reindex when it does not.
Scaling & performance
What scales horizontally, what does not, the knobs that matter, and measured behavior.
Sandbox pool sizing
Warm pool economics, lease pressure, and the reset-or-discard rule.
Task scheduler operations
Settings, metrics, lag alerts, handoff repair, and authority failures.
Observability
Metrics, logs, traces, dashboards, and the signals that predict an incident.
Backup & disaster recovery
What to back up, what is disposable, and the restore order that works.
Troubleshooting
Symptom, check, fix — for the failures that actually happen.
Limits & quotas
Every server-side bound in one table.