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

Acontext

Acontext is an open-source skill-memory layer for AI agents that distills run outcomes into human-readable Markdown files that agents can retrieve and people can inspect, edit and share.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Acontext — 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 Acontext when you want knowledge distilled from agent actions and task outcomes into editable guidance for later runs, rather than only storing conversation facts or retrieving chunks from ingested documents.

Who it is for

  • Developers building AI agents that need to reuse procedures or warnings from earlier runs.
  • Teams that want agent memory in editable, portable files shared across models or frameworks.
  • Agent builders integrating session storage, task tracking or persistent skills through an API and SDKs.

Documented uses

  • Capture what worked or failed in an agent task and turn it into reusable skill guidance.
  • Define a skill-file structure and inspect or correct the resulting memory.
  • Let an agent fetch relevant skill content during a later run.
  • Share or export skill files for use across agents and frameworks.
  • Store agent sessions and messages through the REST API or client SDKs.
Before you choose

Requirements & limits

  • Memory is represented as Markdown files rather than embeddings or a graph, according to the project documentation.
  • Retrieval is described as agent tool use for fetching whole skill units, not semantic top-k search.
  • The project distinguishes persistent learned skills from session-scoped messages, files and artifacts, and from task tracking and session analytics.
  • The repository documents Python and TypeScript SDKs; API docs list OpenAI, Anthropic, Gemini and native message formats.
  • Self-hosting requires Docker, an OpenAI API key and an LLM with tool-calling capability.

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

Start a self-hosted Acontext backend

Create a local Acontext backend and access its API and dashboard.

Before you start

  • Docker installed
  • An OpenAI API key
  • An LLM capable of tool calling

Steps

  1. Download the Acontext CLI with the command below.

    Documented command
    curl -fsSL https://install.acontext.io | sh

  2. Create and enter a working directory with mkdir acontext_server && cd acontext_server.

  3. Run acontext server up.

  4. Open the API at http://localhost:8029/api/v1 or the dashboard at http://localhost:3000/.

Expected outcome

A local backend is available at the documented API and dashboard URLs; the command creates or uses `.env` and `config.yaml` and creates a `db` folder for persistence.

Guide limitations

  • The README describes this path as a quick proof of concept.
  • The LLM must be able to call tools; the README states GPT-4.1 is the default.
Documentation 1 ↗
Source-based practical guide

Create a session and store a message with the Python SDK

Initialize an Acontext client, create a session and store a message using the REST API's documented Python SDK flow.

Before you start

  • Python environment
  • Acontext API key
  • Acontext Python SDK installed with pip install acontext

Steps

  1. Initialize AcontextClient with an API key, such as one read from ACONTEXT_API_KEY.

  2. Create a session with client.sessions.create().

  3. Store a message with client.sessions.store_message(session.id, blob={'role': 'user', 'content': 'How do I reset my password?'}).

Expected outcome

A session is created and a user message is stored through the Python client.

Guide limitations

  • All API requests require a Bearer token; keep API keys out of client-side code.
  • The example stores a message; it does not describe a complete agent-run or skill-learning workflow.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗Guide documentation ↗
Source record details
Source availability
Active
Sources checked
10 October 2026
Original website title
memodb-io/Acontext · Official homepage title · retained in its source language
How we review sourcesSuggest a correction