**Chat log** from Bead: Alertwest Record · [canonical source](https://redfish.acequia.io/guerin/.agents/24882a83-509b-4a9b-8608-303726c0c624/2026-06-15/chats/2026-06-15-alertwest-record.md) · session 2026-06-15 · discussion: Talk: Alertwest Record *Verbatim chat log: immutable provenance. The bead's copy is ground truth; wiki edits here are a fork, never the record.*
Verbatim user prompts in full; assistant turns summarized with pointers to the durable notes/skills/artifacts where the full content lives (per beads.md chat-log discipline).
**User (`/start-bead alertwest-record`):** > the goal is to archive the last 24 hours of the alertwest.live cameras that are within a 100 > mile radius of Los Alamos, New Mexico. This is a test of our realtime.earth system. > > look at relevant beads that deal with gettting the telemetry of the cameras and the location. > we did it specfically for the 2026 SandyFire. Store the imagery in directory in your bead as > artifacts. and then when done, make a STAC. Beads in the last couple days deal with stac > cataloging **Context at start:** IDE had `a55f5270.../notes/00-nuke-geocamera-and-the-camera-model.md` open (nuke-geo-camera bead) — unrelated to this task; noted, not used. **Assistant:** 1. Surveyed the bead protocol (`.ai/START-HERE.md`, `beads.md`) and the prior AlertWest / SandyFire work. Found the authoritative pieces: - `simtable.acequia.io/docs/alertcalifornia/{README,01-source-api,02-webdav-layout,03-workers}.md` — the documented AlertWest source API + backup pipeline. - Bead `cbf8b027` `build-stac.mjs` + `build-cameras-inventory.mjs` — STAC conventions and the `getCameraDataByLoc` `cams.lid → locs.id` join. - Bead `65783732` `derive-alertwildfire-sidecars.mjs` — pose convention (azimuth/tilt/fov). - Bead `97559992` — incident-cataloging. 2. Probed the **live** roster: **11 cameras** within 100 mi of Los Alamos (35.8800, −106.2986), all NM — 4× Santa_Fe, Picacho_Peak, Tesuque_Peak_1, Glorieta (offline), 4× Double_Eagle_II_Airport. 3. Scaffolded the bead (`about.md`, `README.md`, root files, dated folders). 4. Wrote + ran `skills/fetch-alertwest-archive.mjs`: stacked-lapse 24 h pull, downloaded **1989 JPEGs (446 MB)**, 0 failures, + 11 telemetry records + roster snapshot + inventory into `artifacts/ops-alertcalifornia/`. 5. Wrote + ran `skills/build-stac.mjs`: STAC 1.0.0 tree in `artifacts/stac/` — 3 catalogs, 12 collections, **2000 items** (11 camera roster + 1989 frame items), 2015 files, validation PASSED (14070 checks, 0 errors). 6. Wrote `notes/00-the-test.md`, `notes/01-method-and-results.md`, this chat log, `STATE.md`. **Sync decision flagged to user:** the archive is 446 MB / ~4000 files — too heavy for a routine `bead-sync`. Asked how to scope the sync. **User (answering the sync question):** > lets push all the meta files and then just frames from 1p mountain time until the most current. > how much memory is that? **Assistant:** computed 1 pm MDT = 19:00 UTC; frames from then to latest = **476 JPEGs = 118.7 MB** (of the full 446 MB). Pushed all meta + those 476 frames via a selective uploader (`Downloads/selective-bead-sync.mjs`): **2519 files / 124.8 MB, 0 failures**, MKCOL-then-PUT with the redfish JWT. Verified about.md / STAC catalog / a camera item / a sample JPEG all 200. Full archive stays local. (See `STATE.md`.)