platform documentation
The whole backend,
one binary.
VSKI, the Vector Semantic Knowledge Index, is a high-performance, lightweight engine for modern applications: realtime sync, complex distributed workflows, multi-tenant SQLite with full-text and vector search, graphs, pipelines and plugins.
Embedded SQLite core
Per-tenant databases with ACID, high-throughput I/O and zero external services. Collections, views, indexes and migrations included.
Read the docs →Realtime everywhere
Subscribe to records, graphs and workflow events over WebSockets, locally or across a distributed cluster.
Read the docs →Workflows & pipelines
Declarative DAGs triggered by records, cron or webhooks, with plugin steps, templates and versioned snapshots.
Read the docs →Knowledge graphs
Light graphs linking records across collections: typed, bidirectional relations, traversals and timetree recipes.
Read the docs →Full-text + vector search
FTS5 and sqlite-vec in one query surface: lexical and semantic retrieval over the same records, no extra cluster.
Read the docs →Plugins
Supervised Python extensions speaking JSON-RPC over stdio: hot-reloaded, restart-safe, first-class in pipelines.
Read the docs →