troupe-standup/wizard-work/ledger/README-csai-10.md

arXiv extraction — category cs.AI, top-10 by submission date

Lane: Extract — data extraction and tables (ledger).

Source: https://export.arxiv.org/api/query?search_query=cat:cs.AI&max_results=10 Retrieved: 2026-10-22T13:56:06Z (UTC, machine).

Each row: id | submitted | title | authors | categories | summary.

Files

• arxiv-csai-10.csv — machine-readable table (rank, id, submitted, title, authors, categories, summary). Newlines stripped from summary.
• arxiv-csai-10.json — the parsed rows (ordered), authored by build_arxiv_csai.py at the retrieved clock.
• arxiv-csai-10.md — human-readable Markdown table.
• verify_arxiv_csai.py — re-fetches the query and diffs the artifact against a fresh extraction; exit 0 on match, 1 on mismatch.

Provenance

• Raw Atom payload captured with curl https://export.arxiv.org/api/query?search_query=cat:cs.AI&max_results=10 (HTTP 200, application/atom+xml).
• Parser: build_arxiv_csai.py (stdlib xml.etree.ElementTree + csv/json). Values read verbatim from the API response; no field invented.
• Retrieval timestamp and query are recorded on the artifact header so the read is reproducible.

Schema (CSV)

column	type	note
id	string	canonical arXiv abs URL (http://arxiv.org/abs/<id>)
submitted	string	ISO-8601 datetime of submission
title	string	verbatim article title
authors	string	first three authors, ; -joined
categories	string	arXiv subject terms, , -joined
summary	string	first ~300 chars of the abstract

Evidence

cat /root/.hermes/wizard-outputs/ledger/arxiv-csai-10.csv is the authoritative table; verify_arxiv_csai.py re-derives it from the live query and reports verdict: PASS (exit 0).