User Guides

Detailed, task-oriented guides for using VelesDB and velesdb-memory. See also the docs index for the full documentation map.

Guides are grouped by the surface they document. If you are looking for a specific binding, start with its section — each crate README links here, and each guide links back.

Getting started

Guide	Description
Installation	All installation methods (cargo, binaries, Docker)
Installation options	Decision aid: which install path to pick for core vs. agent memory
Configuration	velesdb.toml configuration reference
Use Cases	Common use cases and recommended configurations
Business Scenarios	End-to-end business problems solved with single queries
API correspondence	Rust, Python, TypeScript and MCP names for open, create, insert, search and recall

Core engine (Rust)

Guide	Description
Core: public API map	Where to find what in velesdb_core — an import map, not a signature reference
Core: collections and metrics	The collection model: distance metrics, payloads, storage modes and their memory trade-offs
Core: VelesQL reference	The SQL-like language across the vector, graph and columnar engines
Core: sparse vectors and fusion	Named sparse vectors (SPLADE, BM25, tag sets) beside dense embeddings, and result fusion
Core: streaming inserts	Bounded-channel ingestion, backpressure and the delta buffer for continuously arriving data
Core: query plan cache	The two-tier LRU plan cache that lets repeated queries skip parsing and planning
Core: Agent Memory SDK (Rust)	The velesdb_core::agent memory subsystems for chatbots, RAG and autonomous agents — the Rust API section of the Agent Memory guide
Core: performance numbers	The measured figures, with their methodology and hardware
Search Modes	What Fast/Balanced/Accurate/Perfect/Adaptive mean and when to pick which
Quantization	Vector compression mechanisms (SQ8, PQ, Binary, RaBitQ): internals, training, persistence
Tuning Guide	The numeric reference: mode defaults, HNSW parameters, quantization trade-offs, memory estimation
Graph Patterns	Graph modeling and MATCH pattern recipes
Multi-Model Queries	Combining vector, graph, and structured data in one VelesQL query
Concurrency & Locking	Concurrent access and file-locking behavior
Write Concurrency	Single-writer-per-collection model, batching patterns, Enterprise tier

Agent memory and MCP

Guide	Description
Agent Memory	AI agent memory: semantic, episodic, procedural, TTL, snapshots
MCP server setup	velesdb-memory: install, every client config, the shared HTTPS daemon, embedding/extraction backends
Context compiler	Deterministic prompt compression: budgets, preservation rules, risk, transcripts, the PostToolUse hook
Extraction models	Picking the local model that turns remembered facts into graph edges: criteria, VRAM tiers, and why schema discipline outranks size
Temporal Memory	Dated recall and reasoning about when things happened

CLI and REPL

Guide	Description
CLI command reference	velesdb: every subcommand and flag, import/export formats, packaging, error reference
REPL reference	velesdb repl: dot-commands, session commands and settings, output formats
VelesQL cookbook (CLI & REPL)	Runnable VelesQL snippets: vector, hybrid, sparse, temporal, graph, aggregation, JOIN

Server

Guide	Description
Server REST tour	velesdb-server: runnable curl recipes for collections, search modes, VelesQL, graph, MATCH, errors
Server deployment	velesdb-server: Docker, Kubernetes probes, rate limiting, CORS, startup update check
Server Security	API keys, TLS, CORS, and operations hardening

Python binding

Guide	Description
Python API reference	Database, Collection, search and storage — the authoritative signatures
Python RAG pipeline	From raw text to search results, end to end
Python agent memory	MemoryService and the Agent Memory SDK from Python
Python context compiler	Token-budgeted, provenance-audited prompt context
Python graphs	Persistent GraphCollection and in-memory GraphStore
Python VelesQL	The VelesQL parser API exposed to Python
Python remote server	Using the Python SDK alongside a remote velesdb-server
Python Performance	Throughput tuning for the Python binding
Engine benchmarks	The native Rust numbers behind the Python bindings

WASM (browser)

Guide	Description
WASM JavaScript API	The full surface exposed by @wiscale/velesdb-wasm
WASM persistence and format	How a browser-side store survives a reload: IndexedDB, binary format, performance
VelesQL in the browser	VelesQL parsed, validated and executed entirely client-side

Node.js addon

Guide	Description
Node.js addon	@wiscale/velesdb-memory-node: every method, the JS-side contracts, the context compiler, the bundled agent
Building the Node addon	Building from source when no prebuilt binary covers your platform

Mobile (iOS / Android)

Guide	Description
Mobile API	The complete velesdb-mobile UniFFI surface, in Swift and Kotlin
Mobile build	Cross-compiled libraries and bindings an Xcode or Gradle project can consume

Tauri plugin

Guide	Description
Tauri plugin reference	tauri-plugin-velesdb: every IPC command, events, permissions, storage modes, error codes
Tauri plugin recipes	tauri-plugin-velesdb: runnable snippets for graph, sparse, indexes, VelesQL, events

Data migration (velesdb-migrate)

Importing from another database or a file dump. Not to be confused with the version migration guides below.

Guide	Description
Migrate: CLI and configuration	Command surface and the full YAML schema
Migrate: source reference	Per-source configuration for every shipped connector
Migrate: operations	Throughput tuning, secret handling and troubleshooting
Migrate: embeddings	Rebuilding a velesdb-memory store against a new embedding model

Version migration guides

Guide	Description
Migration v3.3.0	VelesQL correctness + cross-surface parity release migration guide
Migration v4.0.0	Hardening + API-cleanup release migration guide

Tutorials

Tutorial	Description
Tauri RAG App	Build a desktop RAG application with Tauri
Mini Recommender	Build a product recommendation engine with vector search and metadata filtering