Best software, tested
Each list ranks software by what Argusic measured: whether it installed and launched on a fresh machine, its Argusic Score, then stars. Every verdict links to its evidence, and every list shows when it was tested. The lists cover open source projects today; the same procedure applies to any software Argusic tests.
Last tested . Each list shows its own test window; a project can change after that day.
Measured by Argusic on a fresh machine every time. Every number links to its evidence.
| list | tested projects run | last tested |
|---|---|---|
| Open source AI agent frameworks, installed and run | 153 of 158 (97%) | |
AI agent frameworks, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source AI apps you can run yourself, tested | 167 of 175 (95%) | |
Chat interfaces, RAG apps, and agent builders that Argusic installed and launched on a fresh machine with a stand-in model key. | ||
| Open source API clients, built and run | 106 of 109 (97%) | |
API and HTTP clients, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source API gateways and reverse proxies, tested | 96 of 103 (93%) | |
API gateways and reverse proxies, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source CI/CD tools, installed and run | 70 of 77 (91%) | |
CI/CD tools, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source CMS platforms, installed and run | 52 of 56 (93%) | |
Content management systems, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source LLM gateways, installed and run | 32 of 38 (84%) | |
LLM and AI gateways, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source MCP servers that come up on first run | 217 of 228 (95%) | |
Model Context Protocol servers checked by cloning, installing, and starting each one, with the log behind every verdict. | ||
| Open source VPN and networking tools, installed and run | 32 of 38 (84%) | |
VPN and networking tools, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source backup tools, installed and run | 61 of 67 (91%) | |
Backup tools, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source browser automation tools, installed and run | 64 of 65 (98%) | |
Browser automation tools, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source code editors and IDEs, built and launched | 47 of 48 (98%) | |
Code editors and IDEs, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source coding agents that ran on a fresh machine | 217 of 229 (95%) | |
AI coding agents Argusic installed and started on a clean Linux box, each given a real task against a stand-in model. Logs linked. | ||
| Open source databases, installed and queried | 125 of 136 (92%) | |
Databases you can run yourself, each installed, started and given a real query by Argusic on a clean machine. | ||
| Open source developer CLI tools, proven to build and run | 275 of 292 (94%) | |
Developer command-line tools that Argusic compiled or installed and then ran on a fresh Linux machine, each with a logged session. | ||
| Open source e-commerce platforms, installed and run | 47 of 48 (98%) | |
E-commerce platforms, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source ebook readers and libraries, tested | 6 of 6 (100%) | |
Ebook readers and library managers, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source game engines, built and run | 52 of 57 (91%) | |
Game engines, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source games that built and started on a fresh machine | 18 of 22 (82%) | |
Games Argusic compiled or installed from source and started on a virtual display, or as a headless server. Sessions recorded. | ||
| Open source home automation, installed and run | 75 of 83 (90%) | |
Home automation and smart home tools, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source low-code and no-code platforms, tested | 63 of 69 (91%) | |
Low-code and no-code platforms, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source map servers and tools, installed and run | 26 of 27 (96%) | |
Map servers and mapping tools, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source message queues and streaming, tested | 35 of 39 (90%) | |
Message queues and stream processors, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source music servers, installed and played | 64 of 76 (84%) | |
Self-hosted music servers and players, each installed and brought up by Argusic on a clean machine, with the log. | ||
| Open source observability and logging, installed and run | 83 of 88 (94%) | |
Observability and logging tools, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source office suites, installed and launched | 48 of 51 (94%) | |
Office suites you can install yourself, each one installed and opened by Argusic on a clean machine, with the recording. | ||
| Open source password managers you can self-host, tested | 6 of 6 (100%) | |
Password vaults and secret stores that Argusic installed and launched from source on a clean Linux box. Each links to its session. | ||
| Open source project management tools, installed and run | 50 of 54 (93%) | |
Project and task management tools, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source screen recorders, built and launched | 7 of 12 (58%) | |
Screen recorders and screenshot tools, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source search engines, installed and queried | 55 of 59 (93%) | |
Search engines, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source self-hosted analytics, tested on a clean server | 9 of 9 (100%) | |
Privacy-friendly web analytics you can host yourself, each built and started by Argusic from its repository. Ranked by what happened. | ||
| Open source speech to text and text to speech, tested | 113 of 120 (94%) | |
Speech to text and text to speech tools, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source static site generators, built and served | 55 of 57 (96%) | |
Static site generators, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source uptime monitors and status pages, tested | 16 of 16 (100%) | |
Self-hostable uptime checkers and status pages, each installed and brought up by Argusic. Verdicts link to the session evidence. | ||
| Open source vector databases, installed and queried | 62 of 65 (95%) | |
Vector databases, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source video editors that launched on a fresh machine | 41 of 47 (87%) | |
Video editors Argusic installed on a clean Linux box and opened on a virtual display, plus command-line cutters. Sessions recorded. | ||
| Open source video players, built and launched | 17 of 21 (81%) | |
Video and media players, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source web scraping tools, installed and run | 129 of 133 (97%) | |
Web scraping and crawling tools, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source whiteboards and diagram tools, tested | 43 of 48 (90%) | |
Whiteboards and diagram tools, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source wikis you can self-host, tested | 64 of 66 (97%) | |
Wikis and knowledge bases, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Open source workflow automation, installed and run | 79 of 82 (96%) | |
Workflow automation tools, each installed by Argusic on a clean machine. Verdicts link to the session evidence. | ||
| Self-hosted Notion alternatives, checked by a real install | 11 of 12 (92%) | |
Open source notes, docs, and wiki apps that Argusic installed from scratch and launched. Ranked by that result, not by feature lists. | ||
| Self-hosted dashboards and homepages, installed and run | 22 of 23 (96%) | |
Start pages and service dashboards for home labs, set up by Argusic on a fresh machine and opened in a browser. Sessions recorded. | ||
| Self-hosted git servers that started on a fresh machine | 6 of 7 (86%) | |
Git hosting you run yourself: Argusic cloned, built, and launched each server on a clean Linux box. Each verdict links to its session. | ||
run means the project installed and started on a fresh machine, with its real dependencies or with stand-ins for external services; each list separates the two. Projects still waiting for a test are not counted.
How these lists are made
Every project on these lists went through the same procedure. Argusic starts a clean, single-use machine with nothing installed, hands the agent the subject at a pinned commit, and asks it to make the project work by following the project's own instructions, within a time limit. The session is recorded from the first command to the last, and a verdict is stored only when the full log and the recording arrived.
A project that starts is marked runs; one that starts only after stand-ins replaced external services such as a database or a third-party API is marked runs with mocks; if neither the standard agent nor a stronger one got it running, the verdict is could not verify, never "broken". Each list orders its projects by that verdict, then by Argusic Score, then by how popular the project is on its own source. Opinions, votes, and comments never move a row.
Each list page has its own "How we tested" section written from the runs on that topic, and the full rules are on the methodology page.
Frequently asked questions (FAQs)
What does "run" mean on these lists?
That Argusic installed the project on a fresh machine and started it. A project counts whether it started with its real dependencies or with stand-ins for external services; the list page shows which. A project that could not be started is still listed, at the bottom, with the log of the attempt.
How is each list ordered?
By measurement, not opinion: projects that ran come first, then those that ran with mocks, then those Argusic could not verify. Ties go to the Argusic Score, then to how popular it is on its own source.
Why is a well-known project missing from a list?
Three reasons are possible. It has not been tested yet, in which case it appears on the list unranked as "Not yet tested". It was taken off the list because it does not answer the list's question (a library with nothing to launch, a shell framework, a collection of links) and the reason is stored. Or it was never found by the search terms that build the list; suggestions go to [email protected].
How current are the results?
Every verdict carries the date of its test, and a project can change after that day. New projects are admitted and tested every morning. A finished verdict is not repeated on a schedule; the methodology page explains how to dispute a result.
Where is the evidence?
Every verdict links to the project's Argusic page, where each run has its full log and terminal recording stored with a sha256 fingerprint.
Are the lists sponsored or influenced by votes?
No. Argusic is independent, ranks by measurement only, and publishes every attempt whatever the outcome. Community votes and comments are shown on project pages but never change a rank.
All tested projects: subjects. Figures as of Oct 25, 2026 (UTC).