The following is a list of open source projects I've worked on over the years.
Not everything in the list is necessarily finished, may are just proofs of concept or the minimum viable product.
I build a lot of these things to get an idea out of my head, play with a new tool, or to solve my own problem.
This list excludes pull requests to other open source projects, only my own original repositories are listed unless otherwise noted.
This list is sorted by most recent project first.
Last Updated: January 5, 2025
First Commit: November 20, 2024
A userscript that adds colors and expand/collapse functionality to Bluesky threads. This improves the readability and navigation of threaded conversations by adding visual indicators and interactive controls.
Completed. The userscript is fully functional and available on Greasy Fork.
First Commit: October 16, 2024
A demonstration of voice processing capabilities using WebAssembly, featuring Voice Activity Detection (VAD), Speech-to-Text (STT), Text-to-Speech (TTS), and LLM integration. This project showcases the potential of running these voice processing tasks directly in the browser using WASM, with a focus on offline-first functionality. All processing, including the LLM interactions, can be performed locally without requiring external services.
Completed.
First Commit: December 11, 2024
A web-based management interface for Resonite headless servers running in Docker containers. This application provides real-time monitoring and control of your Resonite worlds through an intuitive web interface. Features include live container status monitoring, world management, user management, server configuration, and console access.
Initial functionality completed. More features may be added later.
First Commit: November 8, 2024
A modern development template for creating custom applications for the Spotify CarThing device. Features TypeScript/SCSS support for the client app, a Deno-based WebSocket server with SOCKS5 proxy support, and automated build/deployment scripts. Includes tools for device proxy setup, port forwarding configuration, and Chrome remote debugging capability.
Completed.
First Commit: October 27, 2024
A simple Deno-based HTTP service that resolves AT Protocol handles (Bluesky usernames) to their DID and PDS (Personal Data Server) endpoints. The service supports both GET and POST requests, is CORS-enabled for browser usage, and can identify if a PDS is the official Bluesky PDS.
Completed. A live version of this service is hosted at resolver.atdev.pro.
First Commit: October 27, 2024
A platform that lets you publish Javascript code to your ATProto Personal Data Server (PDS). The code can then be viewed and executed from any user account directly in their browser.
Proof of concept completed.
First Commit: October 22, 2024
A tool that measures the percentage of image posts on Bluesky that are missing alt text. The project connects to the Bluesky Firehose to receive real-time updates about new posts, analyzing each post containing images to check for the presence of alt text.
Completed. The project is fully functional and running at altometer.atdev.pro.
First Commit: May 4, 2024
A text-driven, turn-based game where you interact with a character named John through natural language. Each turn, you can take any action or say anything to John, and an LLM determines how he responds and how the situation evolves based on your input. The game creates dynamic, emergent storytelling where your choices meaningfully impact John's behavior and the narrative direction.
Proof of concept completed.
First Commit: October 9, 2024
A TUI (Text User Interface) client for the picosky Bluesky client. This provides a terminal-based interface for interacting with the Picosky chat client built on ATProto.
Working with bugs. The client is functional but may have issues due to changes in the psky lexicon.
First Commit: May 1, 2024
A Docker container implementation for llama-cpp-python. This was created to simplify the setup process after encountering difficulties finding and configuring a working container. Based on an example from the llama-cpp-python repository.
Completed. The container is working and ready to use.
First Commit: April 6, 2024
This project presents an alternate reality web browser, powered by an LLM. A search engine provides a list of fictional websites, all of which are generated by the LLM.
Completed. The project is fully functional.
First Commit: April 6, 2024
This userscript was created to improve the experience of selecting samples on Strudel.cc. It reformats the list and adds filtering.
Completed. The project is fully functional.
First Commit: March 26, 2024
This was a proof of concept to create a virtual desktop pet with Godot. This works by making each character/object its own window object.
Completed. The project is fully functional.
First Commit: March 12, 2024
This project was created for bridging Discord data into Cables.gl. A bot client connects to a Discord server and relays message data and metadata to a Websocket server. This data can then be received and displayed in Cables.gl to create interactive experiences.
Completed. The application is functional and is being used by the community.
A sample patch can be found here.
First Commit: January 27, 2024
This server was developed as an accompanyment to the Meta Quest version of HereSphere. I collaborated with the developer to add support for streaming and downloading Youtube videos on standalone VR headsets, as well as the ability to download and manage local content
Completed. The application is functional and is being used by the community.
First Commit: January 6, 2024
This was an experiment to replicate the experience of the act of using a computer resulting in mechanical noise. I was nostalgic for how I used to be able to tell what my machine was doing based on the sounds it was making.
Completed. The proof of concept works.
First Commit: December 29, 2023
This is a tool that captures a screenshot of your Linux desktop at regular intervals and processes it with OCR. The intention is to replicate the functionality of products like Rewind.ai without the need for an LLM or sending all of your activity to an private company.
Completed. The application is running daily on my machine and has come in handy multiple times.
First Commit: December 9, 2023
This application implements a speech to text transcription service for the Neos/Resonite virtual world platform. It also features an implementation of Google Blockly to enable end user to create custom voice commands with a simple graphical programming interface.
Complete. This project is in use by members of the Resonite community to power everything from closed captioning to voice assistants.
First Commit: October 2, 2023
This project was an effort to document the networking protocol used by now defunct creative virtual world platform Anyland. See also the Anyland archive project.
Completed. All of the endpoints have been documented with cURL requests and their associated JSON responses. Additionally a Bruno and Postman collection are available.
First Commit: January 24, 2023
This is a PWA client for the Lemmy federated social network. I started this project because none of the existing Lemmy clients at the time fully replicated the experience I had with the Relay For Reddit client.
Proof of Concept Completed. Logging in works. You can view/save/vote on posts. You can view/vote/create comments. It's mostly complete but needs a lot of polish.
First Commit: October 12, 2023
Lenscript was a system I built to allow me to replicate the natural language state-machine scripting system from the now defunct virtual world platform Anyland. This project allows a developer to create a custom natural language syntax defined using JSON that can be implemented in any environment that supports Javascript.
Completed. The project is fully functional and includes an example for responding to browser interactions using natural language scripting. The documentation still needs to be overhauled.
First Commit: September 8, 2023
This application was built to provide an improved script manager experience in the virtual world platform Overte. It supports starting/stopping scripts as well as bookmarking scripts for ease of access.
Completed. This project is fully functional and is available through the in-world app store.
First Commit: September 18, 2023
This application was built to provide an improved debugging experience in the virtual world platform Overte. It supports keyword and type filtering.
Completed. This project is fully functional and is available through the in-world app store.
First Commit: September 18, 2023
This application was built to provide a simple way for users to get their content into the virtual world platform Overte. Since Overte requires a web server, this project gives you a simple interface for uploading files that can be hosted for free on Glitch.io
Completed. This project is fully functional and is in use by some community members.
First Commit: July 25, 2023
This userscript allows a user to filter out certain keywords from their feed on the Lemmy federated social network.
Completed. This project is fully functional and is in use by myself and others.
First Commit: July 23, 2023
This was a project I built to remove my dependency on Spotify. This pipeline automates the process of downloading local MP3 copies of my most recent reccomendations from Last.FM and Spotify, which is then added to my library for manual playback in Navidrome and my Azuracast radio station.
Completed. This project is fully functional and grabbing new tracks every Monday.
First Commit: February 27, 2023
This is an extension for the Stable Diffusion Automatic1111 WebUI (and derivates) that provides a simple interface for viewing notes for assets in your model library. There's also an accompanying Python script that downlolads all of the metadata for every asset from CivitAI.
Completed. This project is fully functional and is used regularly.
First Commit: March 28, 2023
This is a fork of an existing extension for the Stable Diffusion Automatic1111 WebUI. I included this fork in this list as I made quite a few modifications in the effort of performing weight relationship analysis. This included adding functionality to built a pickle that contains the weight vector for every token in the stable diffusion tokenizer. Additionally you can visually map the relationship between one or more tokens to aid in tweening and fine-tuning embeddings.
Completed. The functionality is present and I had a lot of fun turning cats into rats without prompting.
First Commit: March 27, 2023
This is an extension for the Stable Diffusion Automatic1111 WebUI (and derivates) that provdes a more robust interface for prompt management and sharing. It provides functionality for prompt history and bookmarking, as well as the ability to import and export prompts as JSON.
Completed, but deprecated. The functionality addon is fully functional, but has been broken by newer versions of the A1111 UI. There are alternative extensions now tha that provide this functionality.
First Commit: January 29, 2023
This is a full implementation of a MUD (multi-user-dungeon) implemented with Discord. The project includes a tool to convert CirlceMUD and DikuMUD compatible worlds, NPCs, and items into a modern JSON format for use with DisMUD. Internally entities are built with the BitECS framework to allow for extensible and highly dynamic interactions.
Mostly completed. Everything short of combat has been implemented.
First Commit: November 23, 2022
This is a hosted text to speech service for the Neos/Resonite virtual world platform. The original TTS project that was widely used was hosted on the free Heroku tier by Tessacoil, that tier was removed in late 2022. I developed this fork to take over the project and keep it running.
Complete. The server remains online despite me no longer actively using Neos or Resonite. I intend to keep it online indefinitely.
First Commit: November 5, 2022
This is a locally hosted text to speech system for the Neos/Resonite virtual world platform. This project uses both Google TTS and a OS provided TTS engine. This was built as a private alternative to the hosted solution.
Complete. The application is fully functional and is in use by members of the community.
First Commit: January 11, 2023
This was a Discord bot to monitor the status of the ChatGPT website, which at the time was experiencing intermittent downtime. This was created for the ChatGPT discord server.
Deprecated. ChatGPT no longer suffers from downtime, and the addition of the Cloudflare captcha renders this no longer functional.
First Commit: January 11, 2023
This is a browser extension that bridges the ChatGPT browser interface to a websocket server. The idea was to enable free and easy API access to the ChatGPT model without having to go through the headache of standing up an automated browser instance and trying to bypass the Cloudflare bot protection. At the time there was no API access generally avaialable.
Deprecated. ChatGPT now offers a proper API interface with reasonable pricing.
First Commit: June 22, 2020
This was a helper library to automatically setup controller models and button mappings for the browser WebXR API.
Deprecated. Native mapping support is now much better managed by the WebXR API and supporting libraries.
First Commit: September 2, 2022
This server is meant to provide a full physics engine for the creative virtual world platform Neos/Resonite. The physics engine runs locally and forwards the state of all objects to a Neos rendering client via websockets. There is also a Three.js based viewer implemented for debugging collision geometries.
Incomplete. The server and browser interface are complete. I ran into issues with Neos and its websocket client when it came to the frequency of simulation updates.
First Commit: August 20, 2022
This is a mod for the creative virtual world platform Neos/Resonite. It implements an HTTP web server and a Websockets server into the VR client. This can be used to serve local copy of web apps rather than relying on a third party host and removes the need for a hosted websocket server. This mod is largely targeted at developers who are interested in ensuring the backends for their web-enabled objects in-world can stand the test of time.
Complete. There's also the start of a tool to enable the distribution of complete webapps as in-world VR objects.
First Commit: May 28, 2021
This is an implementation of a QR code reader for use with the virtual reality overlay tool, OVRToolkit. This tool was intended to allow for the spatial anchoring of overlay windows in a virtual environment using a QR code as a reference point.
Incomplete. I worked with the OVRToolkit developer, but we ran into commercial licensing issues with the C++ library we intended to use.
First Commit: October 3, 2023
This is an archive of the now defunct virtual world platform Anyland. This archive includes all public world data available from before the shutdown. There is an ongoing project to recreate the server software allowing for single player exploration.
Complete
First Commit: May 26, 2020
This was a tool for creating a 3D model of your real world environment for use in a virtual world. By tracing an outline of your environment it creates a skeleton that can be used to flesh out a 3D model.
Complete
First Commit: May 6, 2019
This was an attempt at building a clone of the virtual world platform Anyland in WebXR.
Abandoned. I ended up going down a rabbit hole of making GPU shader instanced meshes track with their physics engine counterparts. Moved onto other projects afterward.
First Commit: July 13, 2016
This userscript was created to export a list of portals from Ingress into a CSV, which at the time were a 1:1 match to Pokestops in the recently released Pokemon Go game.
Complete. Not sure if it still works today, but this served its role at the time. There are now more complete services serving this role.
First Commit: October 4, 2020
This project integrated the $P gesture recognition engine with WebXR, allowing for simple training and recognition of physical gestures in XR.
Complete
First Commit: March 14, 2020
This is a wrapper around the OVRToolkit JS API, simplifying the interface and making it better conform to Javascript conventions.
Complete
First Commit: August 6, 2019
This tool intercepted the SteamVR finger tracking data and forwarded it to a websocket, for use in the Anyland virtual world. This data enabled the creation of gesture controls within Anyland.
Complete
First Commit: May 17, 2019
This was a Discord bot that was used to follow the Anyland Discord server from within the Anyland virtual world platform.
Complete. Anyland is now shut down.
First Commit: Feb 13, 2013
This was a NodeJS server that was used to record and playback gesture data from the Leap Motion controller. This was used to develop Webleap at the time we only had a single prerelease developer kit.
Complete
First Commit: July 23, 2014
This was a web extension that allowed for gestured based browser navigation using the Leap Motion controller.
Complete
First Commit: December 8, 2016
This was a project to replace the Pebble Appstore after the company shut down operations. This project eventually went on to become the foundation for the Rebble appstore which is still operating today.
Complete. This has since been migrated to the Rebble organization.
First Commit: November 8, 2014
This was an attempt at creating an AI assistant that could be commanded using natural language via speech to text.
Deprecreated. This was functional, but Wit.ai no longer offers the same API. Now LLMs can do a better job of offering this capability.
First Commit: June 6, 2024
This repository contains custom patches and ops for the Cables.gl standalone Electron app. It includes implementations for WebSocket servers, OpenAI and Groq API clients, yt-dlp video URL grabbing, Youtube search, Google Image search via SerpAPI, SearXNG image search, and a random word generator. Each op comes with documentation and example patches.
Complete. The patches and ops are available for use.