Build edge-to-cloud agent memory with NodeDB
Use NodeDB when an agent system needs one durable, queryable memory store for semantic, graph, document, time-series, and key-value context that can run embedded, offline, or behind a PostgreSQL-compatible server.
npx skills add agentskillexchange/skills --skill build-edge-to-cloud-agent-memory-with-nodedb
NodeDB is a multi-model database and storage engine positioned for AI-agent memory. Use it when an operator or agent developer needs to keep episodic events, semantic vectors, graph relationships, documents, time-series traces, and key-value state in one queryable layer instead of stitching together separate vector, graph, document, and cache services. Invoke this instead of using a product UI normally when the repeatable workflow is building or operating an agent memory and retrieval substrate: run NodeDB locally or through Docker, connect with psql, the ndb CLI, HTTP, WebSocket, RESP, or embedded NodeDB-Lite, model the agent memory collections, and use the upstream AI pattern docs for RAG, GraphRAG, on-device memory, evaluation tracking, and multi-tenant AI SaaS. The scope boundary is agent memory and retrieval storage architecture, not a generic database benchmark, broad PostgreSQL-compatible server, or every NodeDB engine feature.
What this skill actually does
Inputs and prerequisites: NodeDB server or Docker image, ndb or psql client, and an agent or application that needs durable memory, RAG, GraphRAG, or retrieval context.
Setup notes: Run docker run -d -p 6432:6432 -p 6433:6433 -p 6480:6480 -v nodedb-data:/var/lib/nodedb farhansyah/nodedb:latest, or install from source with cargo install nodedb. Then connect with ndb or psql -h localhost -p 6432 and follow the upstream AI pattern guides for agent memory, RAG, GraphRAG, on-device AI, and evaluation tracking.
Source and verification boundary: use https://nodedb.dev/docs as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a Multi-Framework workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.