OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Agent Memory & RAG

Velesdb

VelesDB is a Rust-based local-first data and memory engine for AI agents, combining vector, graph and columnar storage under the VelesQL query language. Its documented memory features include evidence paths for recalled information and a deterministic context compiler.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Velesdb — OpenViST project profile in Agent Memory & RAG.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Consider VelesDB when building an agent or application that needs a local memory and retrieval layer combining vectors, graph relationships and columnar data. The repository offers a Python example for vector search and a separate memory server for MCP clients.

Who it is for

  • Developers building AI agents that need persistent, retrievable memory.
  • Teams combining semantic search, relationship queries and typed data in one data layer.
  • Application builders who want a local memory path and evidence for retrieved facts.

Documented uses

  • Store facts for later retrieval by an AI agent.
  • Search agent data by semantic similarity, graph relationships or structured fields.
  • Inspect the evidence path behind a recalled fact.
  • Compile retrieved context deterministically before passing it to an agent.
Before you choose

Requirements & limits

  • VelesDB combines vector, graph and columnar storage and exposes them through VelesQL.
  • The Python quick-start example uses hand-written vectors; text embedding is an opt-in adapter that downloads all-MiniLM-L6-v2 on first run.
  • The README says collection embedding dimension and similarity metric are fixed at creation; changing them requires a new collection and re-indexing.
  • The project documents a memory server for MCP clients separately from its database and Python example.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Run the documented Python vector-search example

Install VelesDB and run its included example using hand-written vectors.

Before you start

  • Python package installation with pip
  • curl and a Python interpreter

Steps

  1. Install the package with the command below.

    Documented command
    pip install velesdb

  2. Download examples/python/hello_velesdb.py from the repository.

  3. Run the script with the command below.

    Documented command
    python hello_velesdb.py

Expected outcome

The example prints ranked results for the documented “tech” and “tech + music” queries.

Guide limitations

  • This example uses hand-written vectors and does not require a server or embedding model.
  • The embedding dimension and similarity metric are fixed when a collection is created; changing either requires a new collection and re-indexing.
Documentation 1 ↗
Source-based practical guide

Connect the VelesDB memory server to an MCP client

Install the local memory server and register it with Claude Code as an MCP client.

Before you start

  • Cargo for installing the Rust package
  • Claude Code MCP client

Steps

  1. Install the server with the command below.

    Documented command
    cargo install velesdb-memory

  2. Register it with claude mcp add velesdb-memory -- ~/.cargo/bin/velesdb-memory.

Expected outcome

The README provides the commands to install and register the VelesDB memory server with Claude Code.

Guide limitations

  • The documented commands cover installation and client registration; they do not describe a complete agent workflow.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
17 October 2026
Original website title
cyberlife-coder/VelesDB · Official homepage title · retained in its source language
How we review sourcesSuggest a correction