AMBER — 'pops on every screen':
staleTime: Infinity only stops REFETCHING, never the FIRST decrypt. So the
encrypted useSyncedAppData blobs (read-state, preferences) each ran a queryFn on
mount (racing consumers multiplied it) and relay responses landing during
navigation made it look per-page. useSyncedAppData now has interactiveNetwork
(default false): for an interactive signer the blob is LOCAL-ONLY — query not
even enabled, patch() a no-op — no decrypt on mount, no publish on a nav-tab
change. Verified with a mocked x-amber login: network-on => Amber op at startup,
with the fix => 0. (useCrewDismissals/useStash were already gated.)
ROUTING — 'same time for every transport option':
router.project-osrm.org hosts ONE car dataset and ignores the profile in the URL,
so foot/bike/car returned identical numbers. Switched to the FOSSGIS per-profile
servers (routing.openstreetmap.de/routed-{foot,bike,car}); the demo server is now
only a last-resort fallback and its result is flagged approximate ('driving
estimate'). Request steps=true and surface turn-by-turn steps for orientation,
plus a '~X min on a board' skate estimate next to the walking time. Verified:
same trip => walk 13 min / bike 5 / moto+car 2, with 5-9 steps each.
Audit of the newsletter section found two real bugs.
1) /newsletter rendered in an infinite loop (100% CPU, React 'Maximum update
depth exceeded'). Chain: useSyncedAppData returned a fresh object each render
-> useReadState.apply (dep: synced) changed identity -> markNewsletterRead /
markAllRead changed every render -> the page effect (deps include it) re-ran
after every render -> the mark set a NEW state object -> loop. The
notification bell looped the same way while open.
- useLocalStorage: setter is now referentially stable (ref + functional set).
- useSyncedAppData: returned object is memoized.
- useReadState: apply reads synced/interactive from refs (stable); marks are
idempotent (same object when already marked) so React bails.
- NewsletterPage marks read once per visit (mount-once ref guard).
2) Issues + subscription markers were published via the default pool (the
author's own NIP-65 relays) but READ via useSkateRelays (community relay +
reader's relays; logged-out = community relay only), so an admin without
goskatespots in their list would publish drops nobody could see. Composer,
backfill panel and the subscribe hook now publish to getWriteRelays() and
fail loudly if every relay rejects (useSpotChat pattern). NIP.md untouched
(no schema change).
Also: guard the overlay pill/reminder against an empty archive, and drop a
redundant canPublish clause.
On the native app the shell is served from https://localhost, so the relative
/api/mints/known hit the local asset server and returned index.html, surfacing as
'Unexpected token < ... not valid JSON' on /mints.
- Native now fetches https://skate-spots.org/api/mints/known via CapacitorHttp
(bundled core plugin, no WebView CORS), with a plain-fetch fallback.
- Web/dev unchanged (same-origin relative path via the Netlify redirect / Vite
proxy).
- HTML or unparseable bodies now become a friendly 'unavailable' message instead
of a raw JSON parse error.
- Desktop: the nav's overflow-x-auto clips its trailing item when the nav grows,
which is exactly the 'More' menu. Widened the header (max-w-[1700px] vs the
1400px page container), tightened nav padding to px-3 and nowrap. 'More' is
fully visible at 1280/1440/2000px with Marketplace pinned (was missing).
- Mobile: the burger menu hard-coded only Map/Lab/Events/Crews + a few extras, so
Marketplace (and Mints) were unreachable. It now appends every nav-catalog item
not already listed, giving mobile/desktop parity (Marketplace, Mints, …).
RouteDialog was effectively broken — users got an empty dialog / 'no route'.
- Map init: the Leaflet map lives in Radix Dialog content, which mounts AFTER
the first effect pass, so the mapDivRef guard was null, the init effect bailed,
and the map was never created (no route line, ever). Hold the node in state
with a callback ref and add it to the effect deps.
- Height: the global .leaflet-container{height:100%} override beats h-64; against
a height-less parent it resolved to 0 (invisible map). Height now lives on the
wrapper (relative h-64) and the map is h-full.
- Stacking: from the map, Directions rendered the dialog inside the spot Sheet
(two nested Radix modals; the sheet covered the route). The map now owns the
RouteDialog at page level and closes the sheet first.
- Routing transport: call OSRM directly (it sends CORS '*') with the cors proxy
only as fallback; distinguish busy / no-route / unreachable errors and guard
against NaN coordinates.
- Header: shrink-0 logo + whitespace-nowrap + gap, nav shrinks/scrolls so it
can never collide with (or glow over) the SkateSpots wordmark; tighter item
padding. Fixes the top-left logo/nav overlap.
- Amber: useReadState keeps read markers LOCAL-ONLY for interactive signers, so
merely opening the bell/DM no longer publishes (encrypt + sign = 2 popups).
- Reputation: lead with verifiedCounterparties (distinct people), cap each
partner at 3 in the WoT-weighted score, warn when every confirmation is with
one account. NIP.md documents the collusion limit honestly.
- Directions: skip the live GPS call when the cached fix is <2 min old; the
blocking 'Finding your location' overlay only shows with no seed (silent
refine chip otherwise). Near Me also flips the sort to proximity.
- .gitignore: cvm-server/package.json (local dev copy of package.cvm.json).
Owner-only chip (next to the Pro/CVM chips) showing the upstream sats balance,
amber when low. useSasquatchAiStatus gains an 'enabled' option; the panel only
fetches while it is open, so the NIP-98 signature never fires at app startup.
Record the working node, the fund/refund recipe, the aggregator outage, the
'orangesync bad key' trap, and that Netlify snapshots env at deploy (redeploy
after env changes).
If the client asks for a model not in the allowlist, the proxy now uses its
configured default instead of 400-ing. So switching the upstream to a Routstr
node is purely server-env (SASQUATCH_AI_BASE/KEY/MODELS) with no client redeploy.
AGENTS.md documents Routstr discovery (kind 38421, endpoints in 'u' tags), the
live nodes found, and the fund->key recipe.
The proxy now reads its upstream from SASQUATCH_AI_BASE + SASQUATCH_AI_KEY
(falling back to Netlify's injected OPENROUTER_* vars), and the model allowlist
is overridable via SASQUATCH_MODELS. Default behaviour is unchanged (gateway).
This makes switching to a sats-only provider a config change, e.g. Routstr:
SASQUATCH_AI_BASE=https://api.routstr.com/v1
SASQUATCH_AI_KEY=cashuA… (raw token) or sk-… (session)
NOTE: api.routstr.com was returning 404 (empty models) on 2026-09-27, so we are
NOT switching yet. Documented in AGENTS.md.
- netlify/functions/ai.mjs: optional Pro enforcement (SASQUATCH_REQUIRE_PRO=1).
Requires a NIP-98 kind-27235 auth header (x-sasquatch-auth) and, unless the
caller is a platform owner, an active Pro pass (x-sasquatch-pro = a
provider-authored zap receipt) verified OFFLINE — signature, user binding,
platform tag, and paid-amount → tier. Off by default.
- src/lib/proAuth.ts: signs the NIP-98 once per session (no Amber popup per
message) and attaches the Pro receipt.
- sasquatchBrain: threads auth headers through the proxy (streaming + non).
- useSasquatchPro: exposes the winning receipt event.
- Verified: 200 with the gate off, 401 with it on (env reset afterwards).
- AppRouter wraps every route (except /amber-callback and the 404) in ONE
<Route element={<Layout/>}> with an <Outlet/>; the chrome now persists across
navigation instead of being torn down and rebuilt on every page.
- Layout: children optional (falls back to <Outlet/>); the map is full-bleed via
a route check (isFullBleed) instead of a per-page noPadding prop.
- All 59 pages + CraftProfileForm/ErrorState/NewsletterIssueView now render
fragments instead of <Layout> (structure-preserving), imports dropped.
Removes the whole remount class: chrome no longer re-runs its hooks per
navigation, Sasquatch chat state survives navigation, and the header can't
re-open Amber on every page change.
Root cause: Layout is NOT a router layout — ~62 pages wrap themselves in
<Layout>, so its chrome remounts on every route change. The header's
NotificationBell -> useNotificationsRead -> useReadState -> useSyncedAppData
re-decrypts an encrypted kind-30078 event on mount; with Amber (NIP-55) every
nip44.decrypt opens the signer app, so it popped on each menu click.
Fix: for interactive signers, decrypt once per session and serve remounts from
cache (staleTime/gcTime Infinity, refetchOnMount/refetchOnReconnect false) in
useSyncedAppData and useStash. Non-interactive signers refetch as before.
Follow-up (noted in AGENTS.md): hoist Layout into the router so the chrome mounts
once — removes the whole remount class (and SasquatchGuide state resets).
- netlify/functions/ai.mjs: POST /api/ai runs against Netlify's AI Gateway —
Netlify injects OPENROUTER_API_KEY + OPENROUTER_BASE_URL into Functions, so no
key is managed or shipped. Curated model allowlist, best-effort per-IP
throttle, CORS for web + Capacitor, and SSE streamed straight through.
- sasquatchBrain: call the proxy instead of OpenRouter directly; drop
VITE_OPENROUTER_API_KEY entirely; PROXY_MODELS set to the slugs that actually
work via the gateway (its ZDR policy blocks gemini/anthropic/gpt-4o-mini).
- netlify.toml: [functions] directory.
- Verified in production: POST https://skate-spots.org/api/ai -> 200.
Policy: the CVM server queries a single community relay
(wss://goskatespots.nostr1.com). Align .env.example (and the local .env) to it.
The local duplicate unit has been retired; the server runs on one host only.
- scripts/deploy-cvm.sh: backup remote server.ts -> upload -> restart -> confirm
the heartbeat, with --rollback. Host/SSH values come from the never-committed
~/.config/skatespots/cvm.env (or CVM_HOST), so no host details ever enter this
public repo.
- package.json: add `npm run deploy:cvm`.
- AGENTS.md: record that the CVM server runs on ONE external host (deploy via
deploy:cvm) and that the discovery relay, not the server, is the SPOF.
Load protection (public read-only endpoint — protects relays + this box):
- Per-pubkey sliding-window throttle and in-flight cap via inbound middleware.
The transport can't return a JSON-RPC error, so over-budget callers are dropped
and time out (logged).
- Global semaphore caps concurrent upstream relay queries.
- Short-TTL cache keyed by filter set, so repeated identical queries (e.g. a
burst of get_community_stats) don't re-hit the relays.
- All tunable via CVM_RATE_* / CVM_MAX_* / CVM_QUERY_CACHE_MS (documented in
.env.example); maps pruned by a janitor.
CI:
- cvm-server/tsconfig.json type-checks the standalone server; new
typecheck:cvm / check:cvm scripts, with typecheck:cvm wired into npm test so
the server can't rot unnoticed (it was excluded from tsc + eslint).
- MapPage: replace the external 'Open with' menu (Google/OSM/Waze) with the
shared in-app RouteDialog, so the map spot sheet routes natively (incl. the
Transitous transit mode).
- FoodPage: replace the Google Maps handoff with RouteDialog.
- RouteDialog: remove the opt-in external 'Open in' handoff and delete
src/lib/directions.ts — no external map apps anywhere (product decision).
useCVMClient's findCoaches ROLE_KIND, self-tag set and getCommunityStats stopped
at 35012, so Sasquatch could not find the event-organizer (35013) or barber
(35014) profiles we added. Extend LabRole, ROLE_KIND, SELF_TAGS and the stats
queries/counts (lab_total) to cover all Lab kinds 35001-35014.
- sasquatchBrain: new streamSasquatchAI — OpenAI-style SSE via OpenRouter (if
the key is set) then Pollinations; falls back to the non-streaming caller only
when nothing at all was delivered, so blocked-stream environments still answer.
- SasquatchGuide: stream the first pass into a live bubble (raw JSON tool calls
are suppressed so partial JSON never shows; prose-then-tool output is dropped),
finalise the bubble in place, show a blinking cursor, and hide the generic
'thinking...' bubble while a reply streams.
- Drop Pollinations 'mistral' (now a 404) leaving the live 'openai' slug.
- routing.ts: add a 'transit' travel mode via Transitous (MOTIS 2 API, keyless,
open CORS), with a local precision-7 polyline decoder and per-leg parsing.
Road modes stay on keyless OSRM; moto remains ≈car and honestly labelled.
- RouteDialog: transit draws each leg (walk dashed grey, transit coloured by
mode) and summarises depart→arrive, duration, transfers and line chips, with
Transitous/OpenStreetMap attribution.
- directions.ts: opt-in 'Open in' handoff to Google / Apple / OpenStreetMap
(+ Organic Maps via geo: on Android), mode-aware. In-app routing stays the
default; the description notes transit shares the trip with Transitous.
- No scooter/moped profile (deliberate).
- sasquatchBrain: abort each provider request after 15s and give up the whole
multi-provider attempt after 35s, so a hung OpenRouter/Pollinations call can no
longer wedge the chat (it now falls through to the static/local fallbacks).
- SasquatchGuide: send the system prompt as a real system message (both the
first pass and the tool-synthesis pass) instead of smuggling it into a user
turn, and mark tool output as untrusted — blunts prompt injection via event
content.
- useCVMClient: signature-verify (verifyEvent) every relay result, matching the
standalone server and the release-notes claim.
- useCVMServerHealth: stale threshold 1h -> 45m to match the server heartbeat.
- Lab forms: a non-empty email must be valid and website/portfolio must
resolve to an https URL; invalid input shows an inline hint and blocks
publish. Adds isValidEmail() (reused by sanitizeEmail). social/instagram
are exempt since they may be bare @handles.
- New PhotoDropzone component: drag & drop plus multi-select image input,
emitting File[] to the caller. Wired into CreateSpotPage + EditSpotPage
(drop several photos at once) and AddPhotoToSpot, replacing the old
one-file-at-a-time inputs.
- Spot photo uploads run sequentially and dedupe URLs.
- Preserve the geo 'g' tag and paid 'featured' tag when editing any Lab
profile (all 7 craft roles via CraftProfileForm + 5 bespoke pages).
- Render previously collected-but-hidden fields on the Lab profile page:
judge 'events_judged', organizer 'events_hosted', editor hardware/software,
and spot references.
- Include Event Organizer (35013) in Community Stats + counts.
- Cap Title inputs at 100 (was 120) so profiles no longer fail the parser
and silently vanish; MIN_TITLE_LEN 2 -> 1 for 1-char display names.
- Normalize bare-domain URLs (example.com) to https:// before storing instead
of dropping/hiding them (new normalizeUrlInput helper; fixes extra url fields).
- No crew strip for judging/organizing (was wrongly showing Photo & Film);
scrollable Offer Services FAB; live bio counters.
git add -A from the repo root swept in assets/, resources/, bun.lock and
cvm-server locks/.bak. Untrack them + .gitignore; correct AGENTS.md junk list
(cvm-server *source* is tracked). No secrets were involved (only placeholders).
- TradeRecommendationsCard on profiles: shows who recommends this trader
(kind 31873 'Attestor Recommendation') and lets a logged-in viewer recommend
them. New useTraderRecommendations hook.
- OfferDialog sends a best-effort NIP-17 DM to the seller when an offer is made;
ListingOffersPanel DMs the buyer when the seller accepts/declines.
- AGENTS.md notes 31873.
Gate: 101 tests, 0 errors/warnings, boot smoke green.
Promotes 'Offer' from a prefilled DM to a real, queryable event:
- Buyer publishes a 30403 offer (d/a/p + optional amount + message)
- Seller sees incoming offers on the listing detail page and can Accept/Decline
(publishes a 30403 response referencing the offer)
- Buyer sees their offer's status and can Withdraw it
- New: useMarketplaceOffers (parse + hooks), OfferDialog, ListingOffersPanel
- NIP.md documents kind 30403; AGENTS.md updated
Gate: 101 tests, 0 errors/warnings, boot smoke green.
On /trades, each authored trade record gets a 'Revoke' action (with a confirm
dialog) that republishes the 31871 with s=revoked — retracting it from
reputation for a disputed or mistaken deal. The counterparty keeps their own
attestation. Revoked records are already excluded from aggregation.
Gate: 101 tests, 0 errors/warnings, boot smoke green.
- New '/trades' page: action-needed (incoming 'confirm this sale' requests as
seller, pending confirmations as buyer) with inline actions, a reputation
card, and the user's own trade records (role/counterparty/state) linking to
each listing. Nav entry (auth-only) + smoke route.
- DM inbox relays (kind 10050) are now block-filtered on both read and delivery
(closes the last relay caveat).
- AGENTS.md updated.
Gate: 101 tests, 0 errors/warnings, boot smoke green.
- Mints page: 'Online only' toggle + result count ('N of M mints')
- Wallet page: 'Browse Cashu mints' link to /mints
- ListingTradeActions now renders the EscrowPanel for a viewer's active escrow,
so escrow status shows on both the grid card and the detail page
- AGENTS.md project map: marketplace detail page + shared actions
Gate: 101 tests, 0 errors/warnings, boot smoke green.
So Local relays are used and Blocked relays are excluded EVERYWHERE, not just in
the NPool routers:
- useListingReports, useSellerInfo, useSkateRelays (read+write), useZaps
(NIP-57 relays tag), usePurchasePro, CollectionDetailPage, CrewForm,
useDirectMessages write fallback
- AppProvider's empty-relays defaults check is the only intentional direct reader
- AGENTS.md updated
Gate: 101 tests, 0 errors/warnings, boot smoke green.
The last 7 react-refresh warnings are dev-only HMR DX from intentional
component+hook/helper colocations (EventCountdown, SortDropdown, TrustFilter,
LanguageContext, NewsletterPage). Splitting them into new files is broad import
churn for no runtime benefit, so each carries a justified file-level disable.
npm test: 0 errors, 0 warnings, 101 tests; boot smoke green.
- RelayMetadata.dmRelays; Relays UI to add/remove, publishing kind 10050
- NostrSync fetches the user's 10050 on login (alongside 10002)
- DMs deliver gift wraps to recipients' + own DM relays (fallback: write relays)
and read from own DM relays + NIP-65 read relays
- useDmRelays() to look up others' DM relay lists
Also fixes a latent data-loss bug: NostrSync's 10002 fetch replaced
relayMetadata wholesale, which would wipe the new local/blocked/dm lists on
login — it now merges. 'Trusted' category intentionally skipped (no distinct
behaviour -> no empty UI); documented in AGENTS.md.
Add a purposeful subset of Amethyst-style relay categories (not a full clone):
- App-local 'Local relays' (never published; added to reads+writes)
- App-local 'Blocked relays' (excluded from every read/write)
- src/lib/relays.ts: resolveRelays()/useRelays() + URL normalisation
- NostrProvider req/event routers go through resolveRelays (blocklist applied)
- RelayListManager: Local/Blocked editors; NIP-65 save now preserves the new
lists instead of overwriting relayMetadata
Deliberately omitted (documented in AGENTS.md): KeyPackage (10051, no MLS
consumer) and proxy/broadcast/private-home (no such discovery model here).
Follow-up noted: a few hooks still read config.relayMetadata.relays directly.
New '/mints' page: browse Community MintRadar mints with health (online/latency/
degraded/archived), NUT count, terms, and add-to-wallet on native.
MintRadar has no CORS, so it's fetched same-origin at /api/mints/*:
- prod: Netlify proxy redirect in public/_redirects (+ 10m cache in _headers)
- dev: Vite server.proxy in vite.config.ts
This also hides users' IPs from MintRadar. Data is unofficial/rate-limited;
the page degrades to a friendly error state. Nav entry + smoke route added.
Add a step-by-step Android verification checklist to ESCROW.md (preconditions,
advertise key -> buyer escrow -> seller redeem -> refund -> guard rails) and
point to it from the Status section and AGENTS.md. Escrow stays unverified until
every box passes on a device with tiny amounts.
Completes the escrow feature on top of the tested core (src/lib/cashuEscrow.ts):
- useCashuEscrow: native-only hook wiring createEscrow (cashu-ts mint + NWC/WebLN
funding), redeemEscrow, escrow-key derivation (coco keyring) and profile publish
- EscrowKeyCard (Wallet): advertise the kind-0 'cashu_escrow' pubkey
- EscrowRequestDialog (listing card): buyer locks funds (P2PK to seller, refundable
after N days) and sends the 31872 carrying the cashu_token
- EscrowPanel + 'Redeem escrow' in the DM request banner (seller side); the panel
is offline/parse-based so it renders on web too
- Buyer 'Escrow & request' supersedes the plain request when available
EXPERIMENTAL: mint-touching paths are unverified on a device (native-only wallet);
use small amounts first. Gate: 101 tests, 0 errors.
Real, oracle-free escrow built on Cashu P2PK with a locktime + refund key:
buyer locks funds to the seller; seller redeems on delivery; the buyer can
reclaim after the deadline. Locked (not plain) tokens are what make the
cashu_token on a public 31872 safe.
- src/lib/cashuEscrow.ts: pubkey normalisation (x-only -> compressed), P2PK
policy builder, offline token inspection (amount/mint/locktime/refund), and
an injectable createEscrowToken mint flow (quote -> pay -> mintProofs{p2pk})
- 15 unit tests incl. the mint orchestration against a fake wallet
- Protocol: advertise a 'cashu_escrow' pubkey in kind 0 (NIP.md); token rides
the existing 31872 cashu_token tag
- ESCROW.md: design, threat model, status, limitations
- AGENTS.md: project map + docs index updated
Wallet UI (cashu-ts + payment rail + profile publish) is native-only and still
needs on-device verification.
- Add a 'this file is the ONLY persistent memory' directive: record non-obvious
learnings in the same session, fix stale entries immediately, date claims
- Add a 'Project map' (HashRouter, custom kinds, marketplace/wallet/DM/lazy
routes, npm test baselines, untracked-junk list)
- Fix stale facts: BrowserRouter -> HashRouter; /src/pages + /src/lib were
understated; describe the smoke test's hash routes + crash-screen detection
- Strengthen System Prompt Management to treat AGENTS.md as long-term memory
Symptom: in the long-running dev server some lazy routes (/settings, /stash,
/collections, /stats) hit the route error boundary ('This page slammed') while
others worked. Console: 'Failed to fetch dynamically imported module'; network:
504 for node_modules/.vite/deps/*.js. Vite's dep-optimizer cache had gone
missing after the server re-optimized on a lockfile change, leaving stale ?v=
hashes.
- Add lazyWithRetry: on a chunk-load failure, retry once, then do a single
full reload per session (avoids reload loops); clears the flag on success
- Use it for all 53 lazy routes in AppRouter
- Document the dev dep-cache gotcha + fix in AGENTS.md
