The Autonomous Stack
A simple, reproducible score for one specific question — how much you actually control your data and infrastructure.
Not security, not quality, not compliance — autonomy. A formal checklist, not an expert opinion.
Three yes/no questions — can you stop it? can you leave with your data? can you roll back? — determine your Autonomy Level (A0–A3). A separate check on source code openness determines Transparency (T0–T2). Five more questions reveal what the score doesn’t show.
TAS gives you three things: a way to evaluate any technology, a catalog of evaluated tools, and recipes to deploy them.
→ Does your infrastructure pass the TAS test? — 30 seconds, no account needed.
→ Explore the Autonomy Map — 150 technologies, 12 constellations.
What do you need?
I need to evaluate my current stack
Start with the Infrastructure Audit. Eight questions — three structural (can you stop, leave, recover?) and five diagnostic (what holds you inside, where is it heading?). Takes 15 minutes per service.
I need to choose a specific tool
Browse the Technology Catalog. 150 technologies rated by Autonomy (A0–A3) and Transparency (T0–T2).
I need to build from scratch
Follow the Minimal Autonomous Server recipe. Or see the Recommended Stack for a curated set of tools across every category.
I need a private phone
Start with the Mobile Apps page. GrapheneOS, F-Droid, and 26 evaluated apps — from messaging to maps.
I need private AI
Start with the AI Tools page. Local inference, open-weight models, and sovereign AI — 15+ tools evaluated across 6 categories.
I want to understand the philosophy
Read whose.world. Or see the Philosophy page for a short bridge.
Core principle
If your system requires an external account to function — you don’t control it.
A tool can be excellent and score A0. A low score is not a criticism — it is a measurement.
Can you break the model?
Every TAS rating should be reproducible: two people, same criteria, same result. We published Assessment Criteria v1.1 with formal, testable checks for every score. Four pilot assessments show exactly how we applied them.
→ Validate the Model — pick a technology, evaluate it independently, tell us if our scores hold up.
License
- Code: MIT
- Documentation: CC BY-SA 4.0