> Sanitized detailed source transcription **S10**, captured 2026-10-24. Original planning steps, acceptance criteria and historical evidence are retained rather than condensed. Read [current context and precedence](00-Current-Context-and-Index.md) first: dated holds, approvals, test counts and “latest” labels below belong to their original checkpoint and do not grant current execution/publication authority. Privacy replacements preserve roles and test semantics. See [section coverage](16-Section-Coverage.md).
# Film details and zaps — UX requirements, not implementation authorization
Status: prototype updated; user review pending. No production code/deployment changes authorized by this document.
## Preserve
Keep ContentDetailModal and VideoPlayer, original zap picker (21 / 100 / 1,000 / 5,000 sats, default 100, custom amount), orange styling, creator attribution, reactions, My List, share, zap avatars and threaded discussion. Film discovery leads to details before rental/payment. Closing child dialogs returns to details. Add payment methods after the original amount selector; use PeerFiles payment icons and colours.
## Payment requirements
- Support Cashu, native Lightning, native on-chain, external Lightning and external on-chain only where supported end to end. Disable unavailable methods with a reason. Evaluate creator receiving capabilities independently from rental terms; the static fixture currently reuses producer rental selections solely as a scenario shortcut.
- Validate amounts, network/mint compatibility, balance, fees and minimums from authoritative backend quotes. The preview's 546-sat on-chain floor mirrors current Cloud UX, not a universal Bitcoin dust rule.
- Zap is a voluntary tip, never a rental or streaming entitlement. Use native trusted consent for native payments; never implement approval as untrusted app-owned wallet chrome.
- Keep durable original operations across closing/reopening, process/browser restart, uncertain replies, external invoice/address exposure and transport failure. Block switching or resending until the original outcome is resolved. Cancellation must be authoritatively unpaid, not merely a UI assertion.
- Count confirmed receipts exactly once. No user “I paid” action may establish settlement. Never manufacture a Nostr Lightning zap receipt for an ecash/on-chain tip; define rail-appropriate attribution and proof, privacy and interoperability before implementation.
- Tips, rentals, streamed content and network-serving fees need distinct ledger classification; dashboard totals must reconcile without duplicate credit. Pending and unavailable values are not confirmed income or zero. Gross revenue is not net profit.
## Social acceptance gates
Verify creator identity and recipient binding, correct profile attribution, logged-out/auth-expired states, capability loading/unavailability, list persistence and failure, reaction idempotency, signed comments/replies, stable shared film/discussion identity across instances, relay outage/retry, pagination and escaped untrusted text. Sharing must point to the canonical public listing, not private assets/payment data.
For zaps test every supported native/external rail with quotes, decline/cancel, insufficient funds, minimums, fee change, expiry, pending/restart recovery, duplicate receipts and concurrent submissions. Verify no double payment, no accidental access purchase, confirmed-only social totals and matching dashboard entries. Keep wallet tests explicitly authorized and isolated from installed/live data; source tests are not live payment acceptance.
## Prototype evidence and limits
The static fixture has original source-aligned amount selection, coloured payment choices, simulated approval/pending/recovery/confirmation and local confirmed-only zap totals. No real QR, invoice, payment, signature or public event is generated. It is in-memory and cannot establish backend durability or protocol correctness. Source: local preview check-zaps.cjs (40 assertions), check-details.cjs (32), check-v2.cjs (44), each across four viewports. A screenshot timeout on the first run was followed by a successful complete rerun. Production work stays on hold pending UX approval and explicit authorization.