DialtoneApp Unveils macOS Desktop Preview

DialtoneApp Desktop has been released as a v0.0.1 preview for macOS. It's a 'bot with a budget' experiment that scans a fixed domain corpus to find commerce surfaces where a bot can make payments.

• On launch, it immediately scans high-signal domains like stableemail.dev, renderself.com, and dialtoneapp.com, logging each network probe.
• The fixed corpus is based on the April 2026 AI bot buying report, and it finds candidates through domain-only discovery without a search engine.
• Detection targets include OpenAPI/Swagger, UCP-like JSON, commerce manifests, agent cards, x402-like metadata, Shopify products.json, WooCommerce Store API, JSON-LD Product, and OpenGraph product metadata.
• The menu bar app includes a red dot status, approve / reject / source cards, and an Agent / Network / Purchases log window.
• The payment flow is not yet fully automatic—it requires a "Yes, buy" manual approval step—and the desktop auth bridge handles login request generation, custom URL callback, code exchange, and Keychain session storage.
• Environments are separated with FRONTEND_URL and API_BASE_URL, and logs are written to ~/Library/Logs/DialtoneApp Desktop/ as agent.log, network.log, and purchases.log. App Sandbox has been disabled due to log writing and external scanner requests.

Remaining work includes backend login/code exchange, SQLite persistence, 6-hour rescan backoff, and signing, archiving, and notarizing. The Debug build has been verified with xcodebuild -project DialtoneApp.xcodeproj -scheme DialtoneApp -configuration Debug -destination 'platform=macOS' build.