Just put Claude to work again to create files that are easier to use for a database import, I really hope this contributes something to the forum and saves you time and work.
I'll think about how we can use more of this to build on the database if you're up for it, if there are ideas just let me know. I had been thinking about seeing how far we can get in trying to restore posthumous books using original sources, searching for gaps like which magazine appearances are still needed, creating an overview of unpublished work, etc
Claude's output: Generated import files cleanly — 407 works, all filenames resolved against the repo. Let me verify the output looks right before handing it over.Both contribution-ready files are in your outputs folder, keyed to `work_id` so they map straight onto the repo:
**`wormwood_uab_links_by_workid.csv`** — the machine-readable master. One row per Wormwood poem: `work_id`, title, the actual repo filename (e.g. `100-jumbo-gem-clips-and-some-hollywood-cigarettes-112.md`), issue, page, and both the page-scan and whole-issue UAB URLs. All 407 works, every filename resolved against the repo, all but one with an exact page link. This is what the moderator scripts a batch import from.
**`wormwood_external_links_snippets.md`** — the human/PR-ready version. One block per work file, each showing the exact `## External Links` line(s) to drop in, with a header explaining placement (after `{{ notes_section() }}`, before `{{ appearances_section() }}`). A reviewer can paste these straight into a pull request, or the moderator can eyeball them.
So the full set for the forum now is: the **spreadsheet** for discussion and insight, and these **two files** for actually getting the links into the database. They line up — same 407 poems, same UAB scans.
Two honest notes:
- The link label I used is `Scan of magazine appearance (The Wormwood Review #144, p.147)` — matching the database's existing phrasing but adding issue and page. If the moderator prefers the barer house style (just the magazine name), that's a one-line change and I can regenerate.
- This covers **Wormwood only**. LA Free Press and NOLA can't get the same treatment yet — their work files would need page-level JSTOR URLs, and the export only gave issue/year. If you can produce a JSTOR export with page references, I'll build the identical pair for those two.
as long as we're using AI for tasks like building database entries and keep it far away from creating poetry we'll be all right...
