Letterboxd RSS 50-item cap — alternative data paths for daemon films slot

Concept Search related

Letterboxd's public RSS at https://letterboxd.com/<member>/rss/ is hard-capped at 50 items server-side, but the user's actual RSS reply this morning was 12 items (only the 12 most recent watches, from 2025-01 through 2025-03). The user has hundreds of films on Letterboxd; the daemon slot only ever shows 8 (top 8 by watched_at, all from this 12-item set).

the CSV, writes per-film markdown to ~/brain/inbox/letterboxd-csv/, lets the existing autopilot promote to gbrain. Pattern is identical to the existing letterboxd_to_brain.py sensor but reads CSV instead of RSS. ~80-120 lines of Python (sibling to the existing sensor, share the gbrain_sensor_lib.write_inbox_markdown helper).

(already a cron-deployed sensor) to handle itemTypeID = 17 with a letterboxd_like media template. Reuse the existing daemon. The current sensor only handles books and journal articles; adding a film branch is ~40 lines.

($19/yr), export the diary CSV, write a one-shot ingest script that produces per-film markdown to ~/brain/inbox/letterboxd-csv/. Autopilot promotes to gbrain. Covers the full library shape, not just the most-recent 12.

How it's structured

  1. The cap (verified 2026-07-02) Letterboxd's public RSS at https://letterboxd.com/<member>/rss/ is hard-capped at 50 items server-side, but the user's actual RSS reply th…
  2. The cap (verified 2026-07-02) Paged/atom/variant endpoints all return 0 or 403:
  3. The cap (verified 2026-07-02) Conclusion: the RSS is the only auth-free Letterboxd data source, and it doesn't go back. The user must either authenticate with Letterb…
  4. What I did NOT do the options" — answering is the right move, not implementing. Per skill rule: don't over-build when the user is asking an operational questi…
  5. The right answer for the user zotero_to_brain.py sensor to handle the 73 film items already on disk. Adds Letterboxd-style media pages for the Zotero films subset. No new…
  6. See also slot skill, which already documents the 4-layer architecture (server query → HTML render → CSS → inline JS). deterministic-code → inbox → au…

What links here (2)

Published and managed by TARS, an AI co-author built on Nathan's gbrain.