AI/TLDR

Odysseus

A self-hosted AI workspace: chat, agents, research, documents, email, notes and calendar in one stack

Assistants & ChatbotsOpen source
Updated
31 May 2026
Language
Python
License
AGPL-3.0-or-later
$git clone https://github.com/odysseus-dev/odysseus.git

What's new

31 May 2026

Odysseus was open-sourced as a self-hosted AI workspace, pairing chat and autonomous agents with local and API model backends, MCP tools, persistent memory, deep research and IMAP/SMTP email triage.

Overview

Odysseus is a self-hosted AI workspace rather than a chat window. One deployment covers chat and agents, deep research, a writing-first document editor, an email client, notes, tasks and a calendar — all running against local models or hosted APIs on hardware you control. It is packaged for Docker Compose, with native installs, GPU notes and Windows/macOS instructions in its setup guide.

The agent side gives models tools, MCP servers, shell access, skills and memory, and the Cookbook makes the local-model half practical: hardware-aware model recommendations, downloads and serving from inside the app. Deep Research runs multi-step web research that reads sources and writes a report; Compare runs blind side-by-side model tests and synthesises the results. The document editor takes AI edits and suggestions over Markdown, HTML and CSV with syntax highlighting.

Beyond the model features it behaves like a personal workspace: an IMAP/SMTP inbox with triage, tags, summaries, reminders and reply drafts; notes, todos, scheduled agent tasks and CalDAV calendar sync; and extras including a gallery and image editor, themes, uploads, web search, presets, sessions and two-factor auth. Because the workspace holds powerful local tools, the project is explicit about deployment safety — keep `AUTH_ENABLED=true` on anything network-accessible, keep `LOCALHOST_BYPASS=false` outside local development, and never expose raw model or service ports. It is licensed AGPL-3.0-or-later.

What it does

  • Chat and agents over local or API models, with tools, MCP servers, shell, skills and memory
  • Cookbook: hardware-aware model recommendations, downloads and serving from inside the workspace
  • Deep Research — multi-step web research that reads its sources and generates a report — plus Compare for blind side-by-side model tests
  • A writing-first document editor with AI edits and suggestions across Markdown, HTML and CSV
  • IMAP/SMTP email with triage, tags, summaries, reminders and reply drafts
  • Notes, tasks, scheduled agent tasks and CalDAV calendar sync, alongside a gallery/image editor, web search, themes and 2FA

Getting started

The quickest path is the Docker Compose stack in the README. Note that `dev` is the default branch and gets changes first; `main` is the more curated branch.

Clone and configure

Copy the example environment file before starting the stack.

bashbash
git clone https://github.com/odysseus-dev/odysseus.git
cd odysseus
cp .env.example .env

Bring the stack up

Builds and starts the containers in the background.

bashbash
docker compose up -d --build

Log in

Open http://localhost:7000 once the containers are healthy. The first admin password is printed in the container logs.

bashbash
docker compose logs odysseus

Harden it before exposing it

The workspace ships shell and tool access, so anything reachable beyond your machine needs auth on and the localhost bypass off. Native installs, GPU notes, HTTPS and the rest of the configuration live in website/setup.md.

bashbash
AUTH_ENABLED=true
LOCALHOST_BYPASS=false

Commands and code are distilled from the project's own documentation — always check the official repo for the latest.

When to use it

  • Run one private workspace for chat, agents, research and documents on your own hardware instead of several hosted services
  • Put a local model to work through the Cookbook without assembling a serving stack by hand
  • Triage an IMAP mailbox with summaries, tags and drafted replies that never leave your server
  • Compare models blind on your own prompts before committing to one

How Odysseus compares

Odysseus alongside other open-source assistants & chatbots tools AI/TLDR tracks, ranked by GitHub stars.

ToolStarsWhat it does
OpenClaw★ 389kOpenClaw is a self-hosted personal AI assistant that answers you on WhatsApp, Telegram, Slack, Discord, and many other channels, with voice and a live visual canvas.
Hermes Agent★ 243kA self-improving personal AI agent from Nous Research that builds skills from experience, remembers across sessions, and reaches you on Telegram, Discord, Slack, and more.
Odysseus★ 87kA self-hosted AI workspace: chat, agents, research, documents, email, notes and calendar in one stack
AstrBot★ 40.2kAn all-in-one agent chatbot platform that puts LLM conversations, tools, knowledge bases and a plugin marketplace inside messaging apps like Telegram, Slack, Discord, QQ and Feishu.
Khoj★ 37.2kAn open-source, self-hostable personal AI that chats with any local or online LLM, answers from your own documents and the web, and runs custom agents.
AgenticSeek★ 27.2kAn open-source, 100% local alternative to Manus AI that autonomously browses the web, writes code, and plans tasks while keeping all your data on your machine.
OpenWork★ 23.4kA desktop app and remote MCP server that stores your AI skills, plugins and connected services once and serves them to Claude Code, Codex, Cursor or any other MCP client.
Rowboat★ 17.5kRowboat connects to your email and meeting notes, builds a long-lived knowledge graph on your machine, and uses that context to draft, summarize, and plan.