Every Tracking Together / EliOS lead source in one place: the single canonical 520-lead corpus, the 295 image lead-sheets in NotebookLM, exact coverage, the real gaps, and the unblock list. Real numbers only — nothing fabricated.
The 520-lead dataset is byte-identical across the repo, the M4 copy, and the Supabase
table (md5 b89b108a… matched both sides this run). No merge was needed — there is exactly
one true lead corpus.
Served behind the secured /elios portal. Fields: name, phone, service, event, status, rep, added, notes.
520 rows mirrored, row-level-security scoped per login. ref jftakowpjkbcqpvtgwgq.
Origin "Leads Master" sheet digitized 2026-06-04. CSV maps 126 rows to 64 Gerald-List images.
Computed directly from leads.json this run. Name and phone are complete; the address column does not exist in the corpus — that is the single largest data gap.
"Eli Leads" notebook (05bf801e) holds 300 sources — 297 images (295 lead-sheets + 2 image_picker) + 2 web pages + 1 markdown — all Google-OCR'd to status=ready. But the reachable CLI path returns only event/service context, NOT the contact rows. 0 sheets were OCR'd to mergeable rows this run.
Where the 520 leads were captured — Anthony's live event sheets. Counts are exact from the corpus.
Exact status distribution across the 520. STOP (18) is suppressed in the portal UI by default.
What is not yet merged, how big it is, and whether it is reachable.
No address column in the corpus. Reachable only by vision-reading the original lead sheets, not via any CLI path.
~167 unique NotebookLM lead-sheets are a separate capture stream from the 520. Contact rows not reachable via CLI.
~135 unique sheets exist only as NotebookLM images. Warm-download fails (lh3 URL returns HTTP 400 even delayed).
12 sheets found locally (a named-but-"absent" list from MAP.md). Vision-readable, not yet digitized.
Folder exists, no file. Paper or another-device sheet. Forward to digitize and merge.
No lead files reachable in Beeper or M4. BlueBubbles REST not yet wired to pull Anthony's iMessage thread.
What is genuinely blocked right now and the exact action that clears each one.
Verified this run on image e016e7ef: source guide returns a prose summary (event = LA County Fair, service = Solar) but abstracts name/phone/address away and takes no prompt arg; fulltext returns only the 224-char lh3 URL; ask returns empty. Contact rows are reachable ONLY by vision-reading the image bytes.
UNBLOCK → vision-read the ~32 local sheets next pass, or export the full image set from NotebookLM to disk then vision-read.
Referenced source surface(s) with no valid session. Cannot fetch any lead data without auth.
UNBLOCK → an authenticated session / credentials for that surface.
The newttleads folder exists but holds no lead file (.DS_Store only). The capture is on paper or another device.
UNBLOCK → Anthony forwards the sheet/photo for digitization → merge into leads.json.
Nothing lead-bearing is reachable from Beeper or M4; Anthony's iMessage is on BlueBubbles, whose REST API is not yet wired.
UNBLOCK → wire BlueBubbles REST, or forward any lead email to east@xlrd.org.
Master unblock for any re-provisioning / migration (per-account export fix, lead→Calendar, color disposition). Not in repo by design.
UNBLOCK → paste from Supabase → Settings → API → service_role when running a migration.