- ui-shared: Add Vite library config and minimal entry point
- cashu: Add TypeScript config and minimal entry point
These packages are placeholders that will be implemented as needed.
The scaffolding allows the monorepo build to complete.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Core Nostr client library for ATM communication:
- Identity management (generateIdentity, load from nsec/hex)
- NIP-44 encryption/decryption for sensitive data
- Event creation utilities (machine status, transactions, auth)
- NostrClient class with relay management
- NIP-42 authentication support for private relays
- Automatic reconnection with exponential backoff
- Subscription management
Types for ATM-specific events:
- Kind 30078: Machine status (replaceable)
- Kind 30079: Transaction records (replaceable)
- Kind 21001-21003: CLINK protocol events
- Kind 22242: NIP-42 auth
Includes unit tests for identity, encryption, and events.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Fix devenv.nix: rename pre-commit to git-hooks (devenv API change)
- Use nixpkgs rust packages directly (avoid rust-overlay dependency)
- Remove non-existent packages (nak, libnfc, httpie)
- Use webkitgtk_4_1 instead of webkitgtk for Tauri compatibility
- Add devenv.lock and pnpm-lock.yaml for reproducible builds
- Update root .gitignore to exclude .claude/ and .pre-commit-config.yaml
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Catalog all existing lamassu-machine drivers (7 validators, 5 dispensers,
3 printers, 4 scanners)
- Add "Port, don't rewrite" principle - leverage battle-tested existing code
- Update HAL structure to include all drivers to port
- Replace NV200 example with ID003 (default validator in lamassu-machine)
- Add Printer trait definition
- Update napi-rs bindings with ValidatorDriver and DispenserDriver enums
- Add configuration mapping for migrating from legacy lamassu-machine config
- Update milestone checklist with prioritized driver porting tasks
- Add driver complexity ranking to guide porting order
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Market context added:
- 95%+ activity is cash-out (selling Bitcoin for cash)
- Typical for remittance corridors, Bitcoin-paid workers
- Bill dispenser is REQUIRED for testing, not optional
Updated development priorities:
- Cash-out first, cash-in second
- Dispenser in minimum dev kit (~$655-855)
- Bill validator becomes secondary priority
Updated milestones:
- Priority 1: Cash-out with real dispenser
- Priority 2: Cash-in (mock validator OK initially)
- Priority 3: Advanced features (Cashu, etc.)
Phase 4 reorganized to implement cash-out before cash-in.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Hardware section now includes:
- One-way dev kit (cash-in only): ~$535-735
- Two-way dev kit (cash-in + cash-out): ~$955-1,555
- Bill dispenser options: Puloon LCDM-1000/2000, Fujitsu F53
- Sourcing tips for used equipment
HAL section now includes:
- BillDispenser trait definition
- Puloon LCDM RS-232 implementation
- napi-rs bindings for dispenser
- TypeScript usage examples for both validator and dispenser
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Comprehensive redesign using Nostr for all ATM communication:
Lightning.Pub as Core Server:
- Replaces lamassu-server entirely
- Nostr-native account system wrapping LND
- Zero server config (no DNS/SSL/ports)
- CLINK native, one-line deployment
- Built-in liquidity management
Private Nostr Relay:
- NIP-42 authenticated relay (strfry/rnostr)
- Whitelist: ATM npubs + Operator npubs only
- Encrypted command/control channel
- Negentropy sync for offline reconciliation
Machine Identity:
- Each ATM has Nostr keypair (nsec/npub)
- Replaces client certificates
- Cryptographic authentication
- No PKI/CA required
Replacing SMS (KYC vector elimination):
- NIP-17 encrypted DMs for receipts
- No phone numbers collected
- User provides npub voluntarily
- Operator alerts via Nostr events
Event Schema:
- Kind 21001-21003: CLINK protocol
- Kind 30078: Machine status (replaceable)
- Kind 30079: Transaction records
- Kind 14: Encrypted receipt DMs
This is the logical extension of CLINK - if we're using
Nostr for payments, why not use it for everything?
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
BOLT12 problems (per @justin_shocknet):
- Onion messages add latency and failure probability
- Every hop is a point of failure (Tor-like)
- Normalizes mobile nodes (bad for Lightning)
- Redundant (LND keysend already exists)
- Astroturfed by NGOs with questionable motives
CLINK Offers advantages:
- Uses Nostr relays (commodity, trustless via NIP-44)
- No HTTP callbacks, WebSockets, or onion messages
- Keys decoupled from Lightning node identity
- Already working: ShockWallet, Lightning.Pub, Stacker News
- More traction in weeks than BOLT12 in years
Updated:
- Protocol stack recommendation
- New components (clink-handler)
- Implementation roadmap
- References section
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Comprehensive rethinking of the project vision:
Core Principles:
- KYC-free: No identity collection
- Lightning-native: Security through protocol
- Self-custodial: Operator and user control keys
- Privacy by default: Minimize data retention
- Autonomous machines: Reduce server dependency
Key Technical Decisions:
- LNbits as primary Lightning backend
- Cashu ecash for offline capability and privacy
- LNURL-withdraw/pay for user-friendly flows
- BOLT12 offers for static payment codes
- NFC BOLT cards for tap-to-withdraw
- Fedimint option for community custody
Architecture Options:
- A: Lean Server (recommended starting point)
- B: Serverless Machine (maximum autonomy)
- C: Fedimint Community Model
Assessment: 60%+ of existing lamassu-server is compliance code.
Recommendation: Rebuild core, reuse hardware drivers.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Documents e-ink options for testing:
- Modos Paper (75Hz, open-source FPGA) - ships Jan 2026
- DASUNG Paperlike 253 (33Hz, frontlight)
- Waveshare development boards
- Industrial/outdoor options (SEEKINK, Geniatech)
Includes honest assessment of trade-offs and recommended
use cases (outdoor, solar-powered, idle signage).
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Comprehensive document covering:
- Compute: Raspberry Pi CM5 (recommended), Pine64 RISC-V (future)
- Bill validators: ITL NV200 with eSSP protocol, MEI Cashflow
- Bill dispensers: Puloon LCDM series, Fujitsu F53/F56
- Displays: Elo I-Series, Waveshare, Faytech open-frame
- Printers: ESC/POS compatible thermal printers
- NFC: ACR122U with libnfc, PN532 modules
- Protocol support matrix and driver migration strategy
- Bill of materials for MVP and full-featured ATMs
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Benefits of shadcn-vue:
- Copy components to repo (you own the code)
- Tailwind-native (no style conflicts)
- Radix Vue primitives (a11y-first)
- Much smaller bundle (~15kb vs ~80kb tree-shaken)
- Full control over styling and behavior
Updated:
- Admin UI: Complete shadcn-vue examples with TanStack Table
- Machine UI: Note about shared shadcn-vue components
- Modernization plan: Updated architecture diagram and bundle sizes
- Bundle target: ~60kb total (down from ~120kb with PrimeVue)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Update architecture diagram to show Vue 3 for both admin and machine
- Replace React decision with Vue 3 unification strategy
- Update machine UI section to include Vue 3 + Tauri
- Add Vue 3 to trade-offs table
- Link to new detailed migration docs
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Migrate from React 18 + MUI to Vue 3 for consistency with machine UI:
- PrimeVue for components (tree-shakeable vs MUI's 300kb)
- tRPC replacing Apollo GraphQL (~5kb vs ~50kb)
- Pinia for state (same as machine UI)
- Shared ui-shared package for common components
Benefits:
- Single framework across all UIs
- ~70% bundle size reduction (400kb → 120kb)
- End-to-end type safety with tRPC
- Shared composables between admin and machine
Includes:
- Complete project structure
- PrimeVue integration with Tailwind
- tRPC client setup
- Passkey authentication flow
- 4-phase migration strategy
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Replaces legacy vanilla JS + jQuery with modern stack:
- Vue 3 with Composition API
- TypeScript throughout
- Pinia for state management
- Vite for building
- Vue I18n for localization
Includes:
- Complete project structure
- Core component examples (App.vue, stores, composables)
- WebSocket integration pattern
- i18n strategy preserving existing translations
- 4-phase migration plan
- Testing patterns with Vitest
- Performance targets (<120kb total bundle)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add lamassu-server, lamassu-machine, lamassu-install as submodules
- Add CLAUDE.md with development guidance
- Add docs/modernization-plan.md with 2026 refactoring roadmap
- Add .gitignore for Node.js/Nix development
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
