Open-source project collections
Analysed projects grouped by what you can realistically build with them — from AI agents and MCP servers to automation, security and self-hosted applications.
Open-Source AI Agents to Build On
AI agents are the most crowded category in open source right now, which makes selection the hard part rather than availability. Dozens of projects implement broadly the same decision loop; what actually separates them is the tool interface, the error handling when a model returns nonsense, and whether you can extend them without forking.
Open-Source MCP Servers and What to Build With Them
The Model Context Protocol turns the N×M integration problem into N+M: a system exposes one server and every compatible AI client can use it. Because the protocol is young, most internal systems, vertical SaaS products and industry databases still have no server at all.
Open-Source LLM and Claude Tooling
The layer around the model has become more interesting than the model itself. Evaluation harnesses, prompt versioning, context management and cost control are where teams currently lose the most time, and where open source is moving fastest.
Open-Source Automation and Workflow Projects
Automation is a crowded category with an unusually forgiving market: buyers already have budget, already feel the pain, and switch tools readily when something fits their process better.
Open-Source Developer Tools Worth Building On
Developer tools are the largest slice of this catalog, and for good reason: developers are the users most willing to adopt something rough if it removes a daily irritation.
Self-hosted Open-Source Applications
Two forces keep this category growing: per-seat SaaS pricing that scales badly with headcount, and data-residency rules that make external hosting awkward or illegal for some buyers.
Open-Source Productivity Projects
Productivity software has an unusual property: users adopt it personally and then drag it into their organisation. That makes it one of the few categories where a good single-user tool has a credible path to a team product.
Open-Source Business Tools
Business tooling is where open source most often meets an actual budget. The buyer already pays for something; the question is whether your version fits their process better or costs less to run.
Open-Source Security Projects
Security tooling has a new and growing driver: a Stanford randomised trial found developers using AI assistance wrote less secure code while feeling more confident about it. Volume of code went up; the review capacity did not.
Open-Source Web Development Projects
Web projects are the easiest to evaluate because you can run them in a browser in minutes, and the hardest to differentiate because everyone else can too.
Open-Source Image and Video AI Projects
Generation quality stopped being the bottleneck here some time ago. What still hurts is everything around it: batching, cost per output, consistency across a series, and getting results into a real production workflow.
Open-Source Audio and Voice Projects
Speech models became cheap enough in 2025 to be boring, which is exactly when a technology becomes useful. Transcription is now a commodity input rather than a product.
Open-Source Data Visualization Projects
Most organisations have more data than questions. The projects in this category are interesting less as chart libraries and more as the front end for data nobody currently looks at.
Open-Source Robotics and Hardware Projects
Hardware-adjacent software has the smallest competitive field in this catalog, because the barrier to entry is physical and most people stop at the simulator.
Open-Source Games and Interactive Projects
Game and simulation code is underrated as a foundation for non-game products. Rendering loops, spatial data structures and real-time state handling transfer directly to visualisation, training and configuration tools.
Open-Source 3D and Architecture Projects
3D tooling serves industries where the incumbent software is expensive, entrenched and widely disliked — a combination that rarely appears elsewhere.
Open-Source Legal Tech Projects
Legal work is document-heavy, deadline-driven and billed by the hour, which makes even a modest time saving directly monetisable.