Odysseus runs natively on Windows without Docker or WSL2. The core features — chat, agent, memory, documents, email, calendar, and deep research — work fully out of the box. The one-command PowerShell launcher handles virtualenv creation, dependency installation, first-time setup, and server start in a single step, and it is safe to re-run at any time.Documentation Index
Fetch the complete documentation index at: https://mintlify.com/pewdiepie-archdaemon/odysseus/llms.txt
Use this file to discover all available pages before exploring further.
Requirements
| Requirement | Notes |
|---|---|
| Python 3.11+ | Install from python.org. During setup, check “Add Python to PATH”. |
| Git for Windows | Provides bash.exe, which is required for Cookbook background model downloads and the agent shell tool. The core app works without it, but Cookbook functionality will be limited. |
One-Command Launcher
Run the launcher
venv\ virtual environment, installs all dependencies, runs first-time setup (which prints a one-time admin password), and starts the server. On subsequent runs it skips any steps already completed.Manual Setup
If you prefer to control each step yourself:Create a virtual environment
py -3.11 is not available, try py -3.12 or py -3.13 — any Python 3.11+ release works.Activate the virtual environment
Set-ExecutionPolicy -Scope CurrentUser RemoteSigned first, or use the launcher’s -ExecutionPolicy Bypass pattern shown above.What Works Natively on Windows
Fully supported natively
- Chat (all API providers)
- Agent with MCP and tools
- Memory and skills
- Documents editor
- Email (IMAP/SMTP)
- Calendar (CalDAV)
- Deep Research
- Compare
- Notes and Tasks
Needs WSL2 or Linux
- vLLM GPU serving
- SGLang GPU serving
http://localhost:11434/v1 in Settings and Ollama handles the rest.Local GPU Models on Windows
vLLM and SGLang require Linux or WSL2. For local model serving on Windows without WSL2, use Ollama:- Download and install Ollama for Windows
- In Odysseus, open Settings and add a new provider endpoint:
http://localhost:11434/v1 - Odysseus will discover available models from Ollama automatically
Troubleshooting
Wrong Python version selected Ifpython on your PATH points to an older interpreter, use py with an explicit version tag for the venv creation step:
py launcher (installed with the Python Windows installer) lets you pick a specific version even if multiple are installed.
.env file encoding
If you edit .env using Notepad, save it as UTF-8 without BOM (in Notepad, choose Save As → Encoding: UTF-8). Odysseus handles BOM-encoded files internally, but saving without BOM avoids any edge-case issues with other tooling.
venv\Scripts\Activate.ps1 is blocked
Run this once in PowerShell to allow local script execution:
-ExecutionPolicy Bypass flag, which does not change your system policy.
Port 7000 is already in use
Pass a different port to the launcher:
APP_PORT=7001 in .env if using Docker.
Next Steps
- Configure models and providers — open Settings after first boot
- Expose Odysseus over HTTPS on your LAN — see HTTPS & Network
- Use Docker instead — see Deploy with Docker Compose