Console: Operate & account

The last two pillars keep your deployment healthy and administered: Operate (Metrics, Storage) and the Account menu (API & SDK, Members, Stacks, Keys & access, Billing).

Operate

Metrics

A live dashboard of the stack’s operational signals, sourced from the same metrics the server exposes at /metrics (Prometheus) and /api/metrics (JSON):

requests — HTTP request counts
storage reads — object-backend read operations
index reads — index-run reads
errors · p95 — error rate and tail latency

Metric labels are deliberately low-cardinality, so the dashboard stays legible even under load.

Storage

Inspect object-storage usage for the stack: how the durable graph, WAL, and index runs map onto objects, CAS write counts, and the byte footprint. This is the window into “object storage is the source of truth” — you can see the runs that back the graph and confirm garbage collection is retiring superseded index runs.

Account

API & SDK

Your stack’s connection details and links to the SDKs. This is where you copy the data-plane base URL and grab install commands for @lbb/client, lbb, and @lbb/mcp, plus the hosted MCP URL for your stack.

Members

Manage who has access to the account — invite teammates and manage their membership. Account sign-in also covers email verification, password reset, MFA, and SSO.

Stacks

Create and manage stacks — each is an isolated tenant with its own graphs, data, and keys. Creating a stack claims a slug (used in MCP URLs) and provisions its storage prefix. Deleting a stack revokes its keys, frees the slug, and purges its data.

Keys & access

Manage per-stack API keys (lbb_sk_live_… / lbb_sk_test_…):

Create keys for a stack.
Rotate a stack key if one may have leaked — rotation issues a new key and invalidates the old one.
Review which keys exist and their scope.

Keep live keys server-side; see Authentication.

Billing

Account billing and plan management. (Billing/quotas are still being built out in the beta — see the overview’s beta note.)

Deleting data

Three levels of deletion exist, mirrored by the API:

Graph reset — a branch-scoped POST /v1/graph/delete?confirm=<graph_id> (also LbbClient.delete_graph) deletes every object under one graph/branch.
Stack deletion — from Stacks, revokes the stack’s keys, frees the slug, and purges tenants/<slug>/….
Account deletion — cascades to every owned stack.

The purge is synchronous today; async/background purge for very large tenants is still in progress.