▐▛███▛█ Claude Code v2.1.316 ▝▜██████▀ deepseek-v4-pro · API Usage Billing ▝▝ ▝▝ ~/tthtlc.github.io

❯ update CLAUDE.md so that all the index pages, and _data/* pages and main page are always updated if any of them are more than 1 day old.

Thought for 4s (ctrl+o to expand)

● I’ll first explore the repo to understand what “index pages”, “_data/*”, and “main page” refer to, so I can write an accurate convention.

Thought for 41s, searched for 4 patterns, read 1 file, listed 3 directories, ran 4 shell commands (ctrl+o to expand)

● I have a clear picture now. Let me read the current CLAUDE.md to edit it precisely.

Thought for 3s, read 1 file (ctrl+o to expand)

● Update(CLAUDE.md) ⎿  Added 26 lines 1 # Project conventions 2
3 +## Index, data, and home pages — freshness rule
4 +
5 +The site’s shell pages and data files must never sit more than +1 day stale. Whenever you work in this repo,
6 +check each file below and bring it up to date before finishing +if its mtime is older than 1 day:
7 +
8 +bash 9 +find index*.html _data -type f -mtime +1 10 +
11 +
12 +Files covered
13 +- Main/home page: index.html
14 +- Index/shell pages: index_all.html, index_gallery.html, i +ndex_search.html, index_spa.html,
15 + index_todo.html, index_discussion.html (and the standalon +e utility index_object_generator.html,
16 + which is not driven by _data)
17 +- _data/*: visualizations.yml, visualization_categories.ym +l, recent_pages.yml, navigation.yml
18 +
19 +How to update each
20 +- _data/visualizations.yml — the master list of every visuali +zation page. Reconcile it against the .html
21 + files actually on disk (add any new page, remove any deleted +page, keep it alphabetically sorted).
22 +- _data/visualization_categories.yml — regenerate: python3 s +cripts/categorize_visualizations.py.
23 +- _data/recent_pages.yml — regenerate: python3 scripts/gener +ate_recent_pages.py.
24 +- _data/navigation.yml — review for stale or broken links.
25 +- Index/shell pages — these are Jekyll templates that read sit +e.data.* at build time, so keeping the data
26 + files fresh is what keeps them current; also verify the hand- +maintained index_todo.html list still points
27 + only at files that exist.
28 +
29 ## Solo graphics content template 30
31 For any *new or modified standalone interactive graphics page * (a single canvas with controls), use

● I’ve updated CLAUDE.md with a new top-level convention. Here’s what it adds:

## Index, data, and home pages — freshness rule

  • A one-line stale-file check: find index*.html _data -type f -mtime +1
  • The exact files covered:
    • Main/home page — index.html
    • Index/shell pages — index_all.html, index_gallery.html, index_search.html, index_spa.html, index_todo.html, index_discussion.html (plus index_object_generator.html noted as a standalone utility not driven by _data)
    • _data/* — visualizations.yml, visualization_categories.yml, recent_pages.yml, navigation.yml
  • How to update each stale file:
    • visualizations.yml → reconcile against files on disk (add/remove/keep alphabetical)
    • visualization_categories.yml → python3 scripts/categorize_visualizations.py
    • recent_pages.yml → python3 scripts/generate_recent_pages.py
    • navigation.yml → review for stale links
    • index pages → kept current by the data files (they read site.data.* at build time), plus verifying the hand-maintained index_todo.html list

One thing worth flagging: the index pages themselves are Jekyll templates that render from site.data.* at build time, so their content is automatically fresh once the _data/* files are regenerated — the convention above reflects that, rather than suggesting a pointless touch on the templates. Let me know if you’d instead want a literal “touch the mtime” step for the index*.html files too, and I’ll adjust.

✻ Churned for 1m 17s · done 1:05 AM