Two features scoped from the stickiness brainstorm, sharing one digest
notification pipeline:
Sadaqah tracker (new tab): a private daily giving journal, not a payment
processor — the app logs, it never moves money. Streak tracking follows
the proven pattern from dedicated apps (Sidq, Daily Sadaqa). Family
visibility is strictly limited to streak counts via a SECURITY DEFINER
RPC (nf_family_sadaqah_streaks) — amounts, causes, and notes never cross
the member boundary, respecting the Islamic preference for giving
privately. Entries can be dedicated 'in memory of' a deceased person from
the Family Tree, with a memorial count (nf_memorial_sadaqah_count, count
only) surfacing on that person's card.
Family Tree social loop: a 'give sadaqah in memory of' link on every
deceased person's card (cross-tab jump via a small requestedTab store in
nav.js), plus completeness hints (missing birth date, missing photo,
no relationships linked) feeding directly into the existing Coverage
Dashboard recommendations engine rather than a new UI surface.
Daily digest Edge Function + pg_cron (07:00 UTC): batches into one email
per family member per day, sent only when there's real content — tree
activity in the last 24h, a birthday/death-anniversary today, or a
weekly sadaqah recap on Sundays. An empty day sends nothing, deliberately
avoiding the notification-spam failure mode the research flagged.
Protected by a shared secret header since it's cron-invoked, not
user-triggered. Verified with a live manual invocation before relying on
the schedule.
Found and fixed two real bugs in nf_family_sadaqah_streaks during E2E
testing: an ambiguous unqualified 'member_id' column reference colliding
with the function's OUT parameter (42702), and a bigint/int type mismatch
from count(*) (42804) — both would have 400'd on every call in production.
Covered by e2e-sadaqah-tree.cjs (12/12). Full regression: 280/280 across
all suites.
Closes 3 of the remaining 4 gaps from the competitive benchmark:
- Coverage Dashboard now has a rule-based recommendations engine
(recommendations.js) surfacing plain-language next steps from data
already in the app — exposed assets, missing Wassiyah, unverified
assets, unlinked liabilities, missing insurance/Zakat setup,
insufficient attestors, no agent assigned, no Waqf configured.
- Faraid Calculator gains a madhab selector (Shafi'i/Hanafi/Maliki/
Hanbali) encoding the one well-documented divergence this engine's
existing rules touch: whether radd extends to a sole-heir spouse
(Hanafi: yes; Shafi'i/Maliki/Hanbali: no, residue unallocated).
Ja'fari (Shia) is honestly gated as unsupported rather than silently
computed with Sunni rules, since it's a structurally different
classification system, not a parameter tweak. faraid.test.js grows
from 14 to 17 cases covering the divergence and the gating.
- Wassiyah tab gains a 'Generate draft will document' button producing
a formatted, statutory-style DRAFT will (declaration/revocation,
executor appointment, bequest schedule, witness attestation blocks)
via browser print-to-PDF — no new PDF dependency. Clearly watermarked
'DRAFT — NOT EXECUTED, requires physical signing and witnessing.'
Not a claim of legal validity, a lawyer-reviewable starting point.
The 4th gap (human-in-the-loop professional tier) remains open — out
of scope for a self-serve prevention tool. COMPETITIVE_BENCHMARK.md
updated to reflect closed/partially-closed status on each item.
Covered by e2e-gaps-round2.cjs (14/14) plus 3 new faraid.test.js cases.
Full regression: 226/226 across all suites.
wmj