DLLM

DLLM · Local desktop chat UI and Python application stack

Build a local chat application for your workflow.

Start with a desktop chat interface, a Python application base, and a reusable template. Add your own commands and LLM-powered features with your preferred coding agent.

To install DLLM Desktop with the Codex setup skill, start a new chat in Codex and paste this prompt:

Install DLLM Desktop using the dllm-desktop-setup skill from [email protected]:dronelabourlm/skill.git.

Once DLLM Desktop is running, create your first application with the Getting Started guide.

Continue to Getting Started

Understand the stack

What is DLLM?

DLLM gives staff with developer skills and individual builders a local desktop shell for personal workflow automations. Its chat UI does not provide intelligence: your registered application does. Start from a reusable application structure, use basic LLM development helpers where you need them, and extend the project with your preferred coding agent.

  • DLLM Desktop provides one local chat UI for your registered applications; it sends messages and commands, but does not decide how the application responds.Explore DLLM Desktop →
  • dronelabour-dllm lets you manage LLM requests in files and call them by prompt name and parameters from Python. Use dllm.connector to switch between provider-specific prompts for the same task and compose calls into pipelines.Explore the Python API →
  • App Hub keeps registered applications up to date and running.
  • chattemplate is the starting project for a conversational application managed by DLLM Desktop. Built on dllm.chatapp in dronelabour-dllm, it provides the application base you can extend with your own commands and message-handling behaviour.Explore ChatTemplate →

Your first working application

From a starter template to your own chat command.

Follow the Japanese Coach tutorial to create a local application from ChatTemplate. Ask Codex to add a server-time command, review and push the change, then restart the application and see it reply in DLLM Desktop.

This first command teaches the development cycle you can reuse as you build your own personal workflow automation.

Build your first DLLM application
Japanese Coach in DLLM Desktop replying with the server time after the custom command is deployed.
The tutorial result: a custom command running in your local application.

Install manually

Prefer to set up DLLM Desktop yourself? Follow these manual installation steps as an alternative to the Codex setup skill.

  1. Install the required software and prepare a repository

    Install Docker, Git, Python, Conda, Visual Studio Code, Node.js, and npm.

    You also need a Git repository for your application, such as one hosted on GitHub or GitLab.

  2. Download and install DLLM Desktop

    git clone [email protected]:dronelabourlm/desktop.git
cd desktop
npm install

  3. Start DLLM Desktop

    npm start

  4. Create your first application

    Click New App +, choose chattemplate, and enter your application's name, description, and Git repository URL.

  5. Extend the application for your workflow

    Follow the Japanese Coach tutorial to add a command with your coding agent, then commit, push, and restart the application in DLLM Desktop.

    Continue to Getting Started