Spectron is a memory and knowledge layer for AI agents – a horizontally scalable application tier in front of SurrealDB, which holds every durable record (graph, vector, document, relational, geospatial) with ACID writes, first-class provenance and trust, graph-resident traces, and tri-temporal belief history. Spectron aims for memory that associates related ideas and keeps straight what was said, what is true now, and what used to be true – much like people do, but queryable and auditable in software.
Use this hub to go from principles to running code, then dive into operational sections (knowledge ingestion, conversational memory, self-hosting, integrations, reference).
Architecture
What Spectron is built to do, what it is not, and how retrieval, traces, and time work:
Welcome and quickstarts
What is Spectron? – product positioning in one pass.
Why agentic memory? – where naive context and pure-vector shortcuts fail.
The accuracy promise – provenance, reconciliation, and auditability.
How it works – end-to-end path from a turn to stored, retrievable state.
Quickstarts
Surrealist dashboard quickstart – create a Context in Surrealist, Playground, Memories, Knowledge, API keys.
Hosted quickstart – Spectron Cloud, API key, first remember and recall.
Self-hosted quickstart – Docker Compose and your first remember/recall calls.
Embedded library – HTTP, MCP, and SDK integration surfaces.
Mental model
How isolation, sessions, categories, and provenance fit together:
Unified substrate and authority – authoritative versus experiential streams in one graph.
Product sections
Knowledge – document ingest, hybrid retrieval, keyword graph.
Memory – sessions, operations (
remember,recall,reflect, …), reasoning model, tuning.Integrations – SDKs, MCP, framework adapters.
Self-hosting – deployment, security, operations, observability.
Cookbooks – end-to-end patterns.
Reference – REST, management API, CLI, configuration, errors.