Configure LLM providers (Google AI, Vertex AI) at the org or project level. Manage API keys, model overrides, and track token usage and costs.
Developer Guide
This guide covers topics for developers integrating with, extending, or operating the Memory platform.
Register external MCP servers (stdio, SSE, HTTP) and the built-in Memory MCP server. Browse the official MCP registry and install community servers.
The Agent2Agent interface: agent cards, project addressing, streaming, and the A2UI surface protocol — with the compatibility matrix of versions and catalogs the server supports.
Monitor and control extraction jobs that turn documents into graph objects. View per-job logs including LLM calls, token counts, and created objects.
Configure Mailgun to enable transactional email (invitations, notifications). Understand the async job queue and delivery tracking.
Health check endpoints, the diagnostics probe, debug stats, and the auto-generated OpenAPI spec.
The public /deploy page: its Railway OAuth routes, the four variables that enable it, and what a 404 from it means.
Complete reference for all 38 permission scopes, umbrella scope expansions, and which scopes are assignable to API tokens.
Reference for memory init, memory ask, memory adk-sessions, memory mcp-guide, and memory install-memory-skills.