# ThesisAnalysisOS > An operating layer for research done with AI agents. Every result is kept with what it stood on, so it can be asked later whether it still holds. It answers holds, fails, stale or unknown, and says why. Python standard library only; it never calls a network or a model. This site is built from the repository's own records, and each page has a Markdown twin. Do not take the site's word: `claims.json` gives every claim with its standing, and `python3 evals/claims.py --json` in the repository reads them back against the code. ## Start here - [This page as Markdown](https://whatstillholds.com/index.md): the idea, a recorded run, the evidence and its limits - [AGENTS.md](https://github.com/JustinBen0131/ThesisAnalysisOS/blob/main/AGENTS.md): how to examine the repository without setting it up, or set it up for a person - [README](https://github.com/JustinBen0131/ThesisAnalysisOS/blob/main/README.md): what it is, in one screen ## Check it - [claims.json](https://whatstillholds.com/claims.json): every claim, its standing at build time, and the command that checks it - [standing.json](https://whatstillholds.com/standing.json): the files this site was built from, by git blob hash - [Evidence as Markdown](https://whatstillholds.com/evidence/index.md): standing of each claim - [docs/CLAIMS.md](https://github.com/JustinBen0131/ThesisAnalysisOS/blob/main/docs/CLAIMS.md): each claim with its code, its test, and what is not claimed - [Recorded measurements](https://whatstillholds.com/records/reuse_frontier.json): the reuse experiment as written by the script ## Understand it - [docs/METHOD.md](https://github.com/JustinBen0131/ThesisAnalysisOS/blob/main/docs/METHOD.md): the failures that led to the design, prior work, predictions - [docs/ARCHITECTURE.md](https://github.com/JustinBen0131/ThesisAnalysisOS/blob/main/docs/ARCHITECTURE.md): a map for reading the code cold - [taos_core/basis.py](https://github.com/JustinBen0131/ThesisAnalysisOS/blob/main/taos_core/basis.py): the one comparison everything else uses ## Use it - [MANUAL.md](https://github.com/JustinBen0131/ThesisAnalysisOS/blob/main/MANUAL.md): daily use - [BOOTSTRAP.md](https://github.com/JustinBen0131/ThesisAnalysisOS/blob/main/BOOTSTRAP.md): the setup interview an agent runs for a person ## Optional - [docs/SPEC.md](https://github.com/JustinBen0131/ThesisAnalysisOS/blob/main/docs/SPEC.md): exact behavior - [docs/LINEAGE.md](https://github.com/JustinBen0131/ThesisAnalysisOS/blob/main/docs/LINEAGE.md): what was tried and did not survive