Agent Memory

Console

Manage projects, keys, memories, evaluations, webhooks, members and settings from the browser.

The console lets you manage your projects and memory from the browser: connection setup, API keys, inspecting memories, a playground, evaluation, webhooks, export/import, members and settings. On the hosted service open https://memory.empecta.com, which takes you to the console at /app/. A local install serves it at http://localhost:5199/app/.

Signing in

EnvironmentHow you sign in
Hosted / productionGoogle, Microsoft (OIDC) or a one-time code sent to your email. The sign-in page shows which of these are enabled.
Local DevelopmentPasswordless developer sign-in; works only from the same machine (loopback).

On first sign-in you complete the onboarding steps (accepting the terms, creating an organization and a first project). Invited users join the organization with an invitation code.

Self-hosted

The console UI is served prebuilt. If you run from source, build it in AgentMemory.App with npm ci && npm run build (or use the development server with npm run dev).

Pages

URL structure: /app/<organization>/<project>/<page>, on the hosted service and on a local install alike.

Setup

Page (menu label)What you do there
InstallPick your client and get ready-made MCP connection settings, plus Claude Code plugin and SDK setup info. You can create an application-specific connection URL.
PlaygroundAdd messages (with extraction or verbatim), run searches, and see why results came back and what was cut by the token budget. This is the only write path that can set an expiry (expiresAt) and tags.
API KeysCreate and revoke keys: application binding, permissions, bound agents, act for users, expiry date. A key is shown once.

Activity

PageWhat you do there
DashboardOverview of the project.
RequestsAPI request log (status, duration) and the change feed.
EntitiesEnd users and entities extracted during consolidation; memories linked to an entity. Edit user profiles.
MemoriesMemory explorer: filtering (application, facet, type …), detail drawer, explain, history, supersession chain, content correction, sharing (tenant admin), deletion, good/bad rating.
EvaluationsSaved recall tests and run history. See Evaluation.
CurationBlocks, manually triggering consolidation, background job status, learning indicators.
GraphRelationship graph between memories, colored by application.
WebhooksCreate subscriptions, test, delivery history, rotate secrets, requeue dead deliveries. See Webhooks.
Export & ImportExport a project as JSON and import it. See Data management.

The Curation, Graph, Webhooks and Export & Import pages depend on your plan including these features; if it doesn't, they appear locked in the menu.

Account and settings

PageContents
Settings → Project → GeneralProject name; LLM provider and your own API key (BYOK), with a connection test.
Settings → Memory ExtractionExtraction and consolidation settings.
Settings → CategoriesCategories: which types get promoted, and extraction instructions.
Settings → RetentionRetention periods.
Settings → OrganizationOrganization name, deleting projects, closing the organization (offboarding).
Settings → MembersInvitations, roles (tenant-admin, member), per-project access.
Settings → PromotionsApprove/reject tenant-wide rule candidates (including approving as a floor).
Settings → Audit LogConsole audit log.
Settings → ProfilePersonal settings; sign out of all your sessions.
Usage & BillingUsage, plan and billing.

Console settings are stored at the workspace or tenant level and override the defaults in the server configuration. For setting keys, see Configuration.

On this page