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

Uteke

Uteke is a local-first memory engine for AI agents, distributed as a Rust binary with CLI commands to store memories and retrieve them through semantic and keyword search. It can also expose an MCP server for compatible agents.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Uteke — 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

Use Uteke when an agent or CLI workflow needs to preserve and retrieve contextual notes locally, including when retrieval should work without cloud access after the initial model download.

Who it is for

  • Developers building agents that need persistent context across sessions.
  • People who want a locally searchable store for agent notes, decisions, or events.
  • Teams using an MCP-compatible agent and seeking a shared memory layer.

Documented uses

  • Save a fact or event with the CLI for later retrieval.
  • Find stored information using a query phrased by meaning or keyword.
  • Configure an MCP-compatible agent to access Uteke through its server command.
Before you choose

Requirements & limits

The project lists release binaries for macOS, Linux, and Windows, Homebrew for macOS Apple Silicon and Linux, Cargo installation, and Docker. Its repository identifies the license as Apache-2.0.

  • The first run downloads an embedding model of approximately 200 MB.
  • The project describes the memory and retrieval workflow as offline, with no API keys or cloud required.
  • Uteke is a memory component; an agent connection requires an MCP-compatible client and configuration.
  • The README reports 98.2% recall@5 on LongMemEval-S for v0.19.0, while strict recall_all@5 is reported as 88.0%.

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

Install Uteke and try local memory search

Install the CLI, store a memory, and retrieve it with a natural-language query.

Before you start

  • A supported macOS, Linux, or Windows environment.
  • Internet access for the first-run download of the approximately 200 MB embedding model.

Steps

  1. Install Uteke using the project’s documented installation options.

  2. Run uteke remember "Deploy v2.1 to staging at 3pm".

  3. Run uteke recall "when do we deploy?" to search by meaning and keyword.

Expected outcome

The CLI stores the example memory and returns relevant results for the recall query.

Guide limitations

  • The embedding model is downloaded on first run; subsequent operation is described as offline.
  • The example demonstrates CLI memory storage and retrieval, not an agent integration.
Documentation 1 ↗
Source-based practical guide

Add Uteke as an MCP server in a project configuration

Provide an MCP-compatible agent with a configured Uteke server command.

Before you start

  • Uteke installed so the uteke-mcp command is available.
  • An MCP-compatible agent that reads the .mcp.json configuration shown in the project README.

Steps

  1. Create or edit the project’s .mcp.json file.

  2. Add the MCP server entry { "mcpServers": { "uteke": { "command": "uteke-mcp" } } }.

Expected outcome

The configuration declares Uteke as an MCP server for the agent to use.

Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
12 October 2026
Original website title
codecoradev/uteke · Official homepage title · retained in its source language
How we review sourcesSuggest a correction