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

Sage

Sage is a self-hosted memory system for AI agents. Connected, authorized agents can save project decisions, failed approaches and lessons to a shared local node, retrieve relevant context in later sessions, and inspect stored experience in the CEREBRUM dashboard.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Sage — 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 Sage for ongoing work where agents need to recall recorded project experience across sessions, rather than relying only on the current conversation or retrieval of source material.

Who it is for

  • People using AI coding assistants on projects that continue across sessions.
  • Teams or developers connecting multiple authorized agents to shared project memory.
  • Developers integrating persistent project memory into an application through the Python SDK or signed REST API.

Documented uses

  • Carry project decisions and lessons into later agent sessions.
  • Record failed approaches so future work can take that experience into account.
  • Share a local memory node across authorized agent clients.
  • Inspect stored experience and manage memory status, confidence and deprecation.
Before you choose

Requirements & limits

The project documents desktop packages for macOS, Windows and Linux, along with Docker and source-build options. It is licensed under Apache-2.0; source builds require Go 1.26.8 or later.

  • Sage is a local memory system, with optional federation, connectors and remote model providers able to send data to configured destinations.
  • Agent enrollment, domain permissions and classification govern access to memories; agent-signed writes record authorship.
  • Memories pass through a validation lifecycle before normal recall, but validation does not prove every remembered claim is true.
  • A personal installation uses one validator; a multi-validator network requires a BFT quorum.
  • The project names Claude Code, Codex and Cursor and describes support for other MCP-capable clients; its README does not demonstrate every possible client.

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

Build and start Sage from source

Build the Sage desktop/server binary and start its setup flow and local service.

Before you start

  • Go 1.26.8 or later
  • A terminal and Git

Steps

  1. Clone the Sage repository and change into its directory.

  2. Build the binary with the command below.

    Documented command
    go build -o sage-gui ./cmd/sage-gui/

  3. Run ./sage-gui setup to choose an AI client and get MCP configuration.

  4. Run ./sage-gui serve to start Sage and the dashboard on port 8080.

Expected outcome

A locally running Sage service and dashboard, with MCP configuration produced by the setup flow.

Guide limitations

  • This covers building and starting Sage; connecting an agent and using memory requires the project’s client-specific connection guidance.
  • The README says recall depends on recorded memories, configuration and agent use of its tools.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
8 October 2026
Original website title
l33tdawg/sage · Official homepage title · retained in its source language
How we review sourcesSuggest a correction