pa.gf.cx equipment-photo batch — 24 iPhone shots → SEO + resize + sort + caption + locate + email cross-ref (parked 2026-05-26)

DARE.CO.UK · PARKED SKETCH · 2026-05-26

Mirrored from ~/.claude/.../memory/parked_sketch_pa_equipment_batch_24_photos_2026-05-26.md. This is a design sketch parked for future build — read for context, not as a current deliverable.

Dan has 24 iPhone equipment photos (~2.4 MB each, ~58 MB total) covering 8 distinct asset classes — pool, tractor, heat-pump, pool-pump, propane tanks (level reading), pool chemistry, dedicated 50A electrical feed for the heat-pump, PenAir / FNS Plus DE filter. Each photo needs SEO-renaming → web-variant resize → schema sorting → caption → location confirmation → final step of cross-referencing emails (Fastmail) + Harvest invoices for receipts. The asset-record photo-drop workflow + the seo-rewrite-for-dropped-images pattern + (likely) a new /pool/ top-level category scaffold are the substrate this leans on. Resume when the certified-water-services A-style redaction commit cycle closes.

Dan 2026-05-26: “I have 24 pictures of equipment, that need sorting into relevant places - pool, tractor, heat-pump, pool-pump, propane tanks (level), pool chemistry, dedicated 50amp power line for heat-pump for electrical, penair cleaner (FNS Plus Diatomaceous Earth filter) - the images, from iphone, average 2.4mb, too big, they need seo-naming, reduction, sorting into logical places in the schema, captions, identify the right location, then finally, look for emails for these items - maybe we have receipts, or bills in harvest.”

The 8 asset classes named

The 6-step per-photo workflow

For each of the 24 photos:

1. SEO-rename

Slug recipe per feedback_seo_rewrite_for_dropped_images.md: - Equipment: <brand>-<model>-<aspect-or-view>.jpg (e.g. pentair-fns-plus-de-filter-rating-plate.jpg) - Tractor: kubota-<model>-front-engine-bay.jpg (whatever the brand turns out to be) - Propane tanks: propane-tank-2026-05-26-level-gauge-reading.jpg (dated because levels are temporal) - Pool chemistry: pool-chemistry-test-kit-strip-results-2026-05-26.jpg

2. Reduce / web variant

Per feedback_asset_record_photo_drop_workflow.md — 2000 px wide @ JPEG quality 82 via sips --resampleWidth 2000 -s formatOptions 82 ... --out *-web.jpg. Important correction from same-day learning: if any photo is already <2400 px wide AND <1 MB, skip the web variant (it would upscale + bloat). At 2.4 MB they’re almost certainly 4032×3024 iPhone landscape, so all 24 likely need web variants. Confirm dimensions before processing. Never sips --rotate iOS share-sheet JPGs (double-rotate trap per feedback_iphone_photo_orientation_handling.md).

3. Sort into the schema

Apply the photo-strip ordering pattern per feedback_photo_strip_order_pattern.md: hero / in-use → details → rating-plate ALWAYS last. Land each photo in the right record folder; if the record doesn’t exist yet, scaffold it from the asset-record template (cf. pa/technology/qnap-ts-453a/index.html for the gold-standard tech-record structure, or pa/contractors/certified-water-services/index.html for the contractor structure).

4. Caption

Each photo gets a descriptive alt sentence (per feedback_seo_rewrite_for_dropped_images.md Part 2) AND a visible figcaption micro-caption in the photo strip. Captions carry brand / model / aspect / cross-reference.

5. Identify the right location (file-system)

• If /pool/ doesn’t exist yet, scaffold it as a top-level category mirroring /well-water/ (per parked_sketch_pa_gfcx_well_water_top_level_category_2026-05-26.md).
• Apply the EOD + stewardship-posture pattern per feedback_tech_record_eod_and_stewardship_posture.md to each pool-equipment record — heat-pumps EOD ~10-15 yrs, DE filters ~15-25 yrs (housing), pumps ~8-12 yrs, etc.
• Cross-reference each record per feedback_cross_references_smart_throughline_connected.md — pool-pump links to heat-pump links to electrical-feed links to chemistry log. The through-line is the pool subsystem.

6. Email + Harvest cross-reference

For each item, search: - Fastmail 3. Projects > Contractors folder + general search for vendor/brand keywords (Pentair, Hayward, Kubota, John Deere, etc.) - Harvest invoice export for purchase + service entries - Receipts in pa/_receipts/ or wherever already-OCR’d receipts live - Amazon orders substrate (already indexed) for any consumables — chemistry test strips, DE powder, chlorine tabs

Result of each cross-reference: a row in the record’s “Files archived” + Job-history sections; vendor link if the record turns into a contractor relationship.

Likely top-level category to scaffold first

pa/pool/ — sibling to /well-water/ — because at least 5 of the 8 asset classes drop under it: - pool (landing) - pool-pump (equipment) - heat-pump (equipment) - chemistry (sub-substrate) - FNS Plus DE filter (equipment)

Plus the electrical-feed and the propane tanks may also cross-reference into /pool/.

The /pool/ landing should mirror the /well-water/ structure: system stack cards (pump · filter · heater · chemistry · electrical · structure · cover · plumbing) + service cadence + vendors-on-file (two-tier) + multi-dropzone (equipment / chemistry-logs / invoices / issues) + test-log archive + open issues + cross-references.

When resuming — first 5 minutes

1. Find where Dan has the 24 photos — likely ~/Desktop/, ~/Desktop/Sort/, or pasted directly into the chat. Confirm path.
2. Triage by asset class — ls -la <path>/*.jpg + Read each to identify which of the 8 classes it belongs to. (Or ask Dan to pre-sort if photos aren’t obviously labelled by filename.)
3. Scaffold /pool/ — if it doesn’t already exist, build the landing page first so the equipment records have a home.
4. Add the dropzone context mapping — ~/bin/pa_dropzone_apply.py needs pa-pool + maybe pa-electrical + pa-equipment (for tractor) mappings before any dropzone use.
5. Process the photos in asset-class batches rather than 1-by-1 — all heat-pump shots together, then all pool-pump shots together, etc. Each batch = one record-build + photo-strip.

Cross-references

• feedback_asset_record_photo_drop_workflow.md — the 5-step photo workflow this multiplies by 24
• feedback_seo_rewrite_for_dropped_images.md — slug-with-purpose + descriptive alt
• feedback_photo_strip_order_pattern.md — hero → details → rating-plate ordering
• feedback_tech_record_eod_and_stewardship_posture.md — EOD horizon + posture rung per equipment record
• feedback_cross_references_smart_throughline_connected.md — connect pool ↔ chemistry ↔ electrical ↔ propane through-line
• feedback_iphone_photo_orientation_handling.md — don’t sips --rotate
• parked_sketch_pa_gfcx_well_water_top_level_category_2026-05-26.md — analog for the /pool/ landing scaffold
• feedback_dropzone_pattern_shipped.md — add pa-pool + sibling contexts to the apply script
• parked_sketch_receipt_vision_ocr_vehicle_history_2026-05-24.md — receipts may already be vision-OCR’d; check before re-extracting
• project_amazon_orders_integration_parked.md — Amazon substrate for chemistry/consumables
• project_harvest_api_integration_parked.md — Harvest substrate for invoice extraction

Resume trigger

• Certified-water-services A-style redaction commit closes
• Dan signals “ok let’s do the 24 photos” or names a specific subset
• The Pentair / FNS Plus + Kubota / John Deere brands get confirmed from the photos (enables vendor-record sketches)

Likely follow-on parked sketches once this lands

• parked_sketch_pa_pool_top_level_category_*.md — the /pool/ scaffold itself
• Per-vendor parked sketches if the equipment owners identify (Pentair, whoever supplied the tractor, the 50A electrical contractor, the propane supplier)
• A parked_sketch_pa_electrical_top_level_category_*.md if /electrical/ doesn’t exist yet