- π Dunoon Daemon App v2.1.0
- βοΈ What is Dunoon Daemon?
- π₯ Core Features
- π₯ Download
- π οΈ Installation
- π€ Models
- π Local & Private by Design
- π‘οΈ Windows Security Notice
- π Documentation
- π§ͺ Release Information
- π» Source Code
- π Credits
- Dedicated to the loyal companions who walk with us through every realm, and the code that keeps their echoes alive.
- βοΈ What is Dunoon Daemon?
π Dunoon Daemon App v2.1.0
Local AI characters with memory, personality, agency β and a world that stays put.
Free β’ Open Source β’ Local-First β’ Windows β’ GGUF
βοΈ What is Dunoon Daemon?
Dunoon Daemon is a free, open-source, local-first Windows application for:
- AI roleplay
- Character interaction
- Companion-style chat
- Tabletop RPG experimentation
- Autonomous multi-character Arena scenarios
It runs GGUF language models locally through its bundled llama.cpp runtime.
No commercial cloud AI API is required for normal operation.
Your conversations, personas and memory remain on your own machine.
π₯ Core Features
| Feature | Description |
|---|---|
| π§ Persistent Personas | Configurable personality, memory and behavioural controls |
| π¬ Solo Chat | Local character conversations using GGUF models |
| βοΈ Arena Mode | Autonomous interaction between multiple AI characters |
| π Scene Authority | Established world facts are tracked instead of casually rewritten |
| π Campaign Lore | Persona-scoped lore and knowledge sources |
| π§Ή Memory Tools | Inspect, manage and purge learned memory |
| ποΈ Situational Awareness | Optional environmental and situational context |
| βοΈ Worldbuilding Controls | Collaborative plot and creative freedom settings |
| π£οΈ Voice Support | Character voice functionality |
| π¨ Skins | Multiple interface themes |
| π¦ Portable Windows Build | Extract and run β no Python installation required |
π₯ Download
Dunoon Daemon v2.1.0
Download:
Release size
974,885,591 bytes
π SHA-256
ZIP
5655A24E4418038271A9DDE0D6DBFCB0311A1997B967CDA4E6DA0E0449B8795A
DunoonDaemon.exe
483AE08AD72F89A37935316EE67790562D13627355601A5E25EA58EB29A37122
These hashes correspond to the tested v2.1.0 release artifact.
π οΈ Installation
- Download
DunoonDaemon_v2.1.0.zip - Right-click the ZIP and choose Extract All
- Keep the entire extracted folder together
- Run
DunoonDaemon.exe
Folder layout
DunoonDaemon_v2.1.0/
β
βββ DunoonDaemon.exe
βββ bin/
βββ data/
βββ models/
βββ Dunoon_Daemon_App_THE_TOME_v2.1.0_PARCHMENT_EDITION.pdf
βββ The_Goblins_Guide_to_Dunoon_Daemon_App_v2.1.0_PARCHMENT_EDITION.pdf
β οΈ Do not separate
DunoonDaemon.exefrom thebinfolder.
β οΈ Do not run the application directly from inside the ZIP archive.
No Python installation is required for the portable release.
π€ Models
GGUF models are not included with Dunoon Daemon.
You can:
- Place compatible GGUF files inside the
modelsfolder - Or select a GGUF model stored elsewhere on your computer
Model quality, memory requirements and context limits vary considerably between models.
π Local & Private by Design
Dunoon Daemon is built around local inference.
The application launches a local llama-server process and communicates with it on your own machine.
Persona data, conversations, memory and application state are stored locally.
No mandatory cloud AI service.
No required API subscription.
Your roleplay does not need to leave your computer.
Users remain responsible for any models, documents, images or other material they choose to use with the application.
π‘οΈ Windows Security Notice
Dunoon Daemon is currently distributed as an unsigned Windows executable.
Because it is unsigned, Windows Smart App Control, Microsoft Defender or other security software may warn about or block the application on some systems.
This is a portable PyInstaller build of the open-source project, not a digitally signed commercial installer.
The SHA-256 hashes above are provided so users can verify the release artifact.
π Documentation
Two manuals are included inside the release archive.
π Dunoon Daemon App: The Tome
The full reference manual.
πΊ The Goblin's Guide to Dunoon Daemon App
A quicker, considerably less solemn introduction to getting the thing running without summoning anything unfortunate.
π§ͺ Release Information
| Version | 2.1.0 |
| Platform | Windows 10 / 11 |
| Model Format | GGUF |
| Inference | Local llama.cpp / llama-server |
| License | MIT |
| Source Checkpoint | ee15b11 |
π» Source Code
The complete open-source project is available on GitHub:
Bug reports, forks, experimentation and community fixes are welcome.
π Credits
sorvan76 (Kepler365) and ChatGPT - a human / AI fusion (2026)